Files
PaperClipAI/server/src/services/native-runtime/external-chat-wait.integration.test.ts
T
DottaandPaperclip 208a093d19 fix(chat): preserve resumed review waits and retry newer saved input (#17725)
<!-- entire-trail-link-start -->
https://entire.com/gh/paperclipai/-/trails/36/changes/paperclip/36
<!-- entire-trail-link-end -->

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - GitHub review agents publish findings and wait for an authorized
reply or head change.
> - Review tools could retain the requester after resume while
completion lost that same wait proof.
> - A yielded report could then queue its own continuation or repeat a
missing-wait error.
> - Ordinary Retry could also select an older failed turn while a newer
saved request had no admission.
> - This change shares the existing authority proofs and canonical
admission owners.
> - Current access, publication, execution and budget gates remain
required.

## Linked Issues or Issue Description

Refs #17318. Builds on the canonical saved-input retry owner in #17733.
Related #16550 covers deferred legacy comments after ownership waits.
The separate recovery repair in #17724 owns the cleanup-to-saved-input
handoff; this PR does not add a second admission owner.

A resumed review could publish an assessment, then turn its
`response_wake` into an immediate `issue_status_changed` wake. A
provider that omitted its continuation could instead repeat a completion
error. A newer linked-human message saved during a budget hold could
also remain unstarted because Retry selected the older failed source.

## What Changed

- Resolve passive review waits through the accepted requester ledger
used by GitHub tools. Require the latest reviewed head, exact company,
task, agent, repository and processed publication under current access.
- Fill omitted/null yielded continuation from that server proof. Permit
an inherited normal tree resume only through its accepted parent
identity. Permit an ordinary follow-up only after this run has a
processed useful reply linked to its exact inbound message. Reject
acknowledgements, partial assessments, pending or uncertain writes as
that reply proof.
- Recheck the same run/review/head/source proof during feedback,
finalization and final status commit. Preserve explicit malformed-input
rejection and Done policy. Add no monitor permission or publication
grant.
- Keep heartbeat and publication projections narrow. Preserve both
published Copilot v62 identities and generate source-bound v63 from the
pinned merged base. No v63 live qualification is claimed.
- Certify the structured isolated-directory binding failure only at its
pre-allocation boundary and only through canonical task Retry. Other
workspace validation reasons and unknown effects remain denied.
- Route Retry to eligible newer saved input through the existing
canonical owner. Preserve the old result and skipped receipt. Require
current principal, source, App, runtime fence, budget, native ownership
and exact cleanup proof.
- Recover absent historical App metadata only through an earlier trusted
processed response receipt and an immutable non-null endpoint claim.
Explicit malformed or conflicting metadata stays denied.
- Revalidate execution through the exact claimed wake, coalesced task
Retry and resolved user-continuation receipt after admission. A second
normal Retry can recover the exact never-started authorization failure
with unchanged saved input. Require empty execution evidence, exact
diagnostics and no allocated or active executor.
- Explain when a recorded result prevents exact replay. Ordinary board
Retry directs a new task message to continue from saved work. Inherited
requester identity alone never grants replay.
- Keep accepted board-message preparation failures with the canonical
task Retry owner. Certify the exact unchanged comment, original user
wake, accepted same-user identity and absence of provider execution. Pin
that source, recheck it at admission, and bind identity initialization
to the coalesced intent and exact claimed successor receipt.
Uncertifiable sources retain an actionable fresh-message refusal.
- Return safe named authority categories for board Retry refusals.
Update execution documentation and capability maps. Add no schema or
polling loop.

## Verification

- A fresh external review request exposed a separate isolated-directory
preparation failure before allocation. Its first normal Retry saved an
intent but admitted no successor. This follow-up certifies only that
exact structured preparation boundary through the existing canonical
Retry owner, with trusted failed settlement and no executor, lease,
result or output. A real HTTP/queue regression reproduces the missing
successor before the repair. Twenty-three current-source safety cases
and the final 98-case Retry/HTTP group pass, including a fresh click
after a denied intent, source revocation, accepted results and the older
direct API remaining denied. The warning case uses actual configuration
preparation and production event append; unknown warning shape or
provider provenance stays denied. Full workspace typecheck, build, maps
and diff checks pass. Renewed hosted normal Retry now succeeds on the
final combined source: it preserves the exact saved request and linked
requester, starts one native successor in a fresh attempt workspace,
publishes the current-head 2/5 assessment and commits In review with
passive response wake.

- A later hosted normal Retry published a low-score assessment but lost
its accepted source identity because the previous attempt failed before
identity initialization. It repeated continuation errors, then submitted
Done. This is failed acceptance, not a successful qualification. The
initializer now retains the exact source only after execution-time
failed-chat Retry authorization; generic Retry and Done policy are
unchanged. A real queue/initializer regression fails with a null source
on the previous code and passes with the newer saved source. Two real
database completion cases verify that source yields the authoritative
wait and commits In review. Renewed hosted normal Retry now retains the
accepted source and exposes the authoritative current-review wait. It
completed after two semantic feedback corrections (wrong contract
revision, then wrong criterion IDs); the correct contract was present in
emitted input. The third finish was accepted and committed In review,
with no missing-wait loop or Done escalation.
- Repeated board-message preparation failures now walk their canonical
intent, failed dispatch and accepted parent identity, rechecking each
contained predecessor. A production queue regression fails the second
ordinary HTTP Retry before the repair and passes afterward. Changed
source, contradictory execution evidence and cyclic receipts remain
denied; traversal stops after twenty links.

- New inherited omission cases failed before the repair. The final full
PostgreSQL 18 wait/status suite passes **252 tests** under Node 24.22.
It covers omitted/null input, malformed-first recovery, same-head
resume, exact-source replies, changed heads, revoked access,
pending/uncertain/mixed publications, final commit revalidation, no
duplicate review and no self-wake.
- **15 existing publication tests passed before the pinned merge** on a
separate fresh database. They retain default partial-delivery and
failure-fallback behavior. Context reads remain below the unchanged 16
KiB heartbeat and publication response limits with large saved output
present.
- Saved-input tests cover a real signed PR webhook, production
accounting-budget refusal, normal policy recovery and canonical Retry.
Historical producer-shaped tests prove the missing-App case red, then
check trusted prior receipts and malformed, late, revoked or mismatched
proof. A broader order-coupled test was reproduced on the prior source;
its global fault injector now targets its own authorization query. The
43-case sequence passes without changing production behavior or its
safety assertions.
- New signed-ingress and real heartbeat execution tests reproduced the
admitted-successor failure, then the next normal Retry failure. The
final pinned merged source passes 249 full GitHub review/publication and
Retry cases on a fresh database, including all 49 saved-input cases. A
separate 74-case Retry/HTTP run also passes. Exact dispatch negatives
cover source, company, agent, task, successor, receipt, pause,
assignment, accepted success, controller and cleanup proof. Bootstrap
contradictions remain denied. The canonical Retry suite passes 16 cases
through its supported external PostgreSQL test connection.
- The normal board HTTP route reproduced the unusable inherited-result
refusal. It now returns actionable guidance without starting a
successor. The actual UI Retry hook and banner tests pass 12 cases.
- A production-shaped accepted board comment followed by a pre-provider
qualification failure reproduced HTTP 409 before the repair. The final
real HTTP/queue test certifies one canonical successor and preserves the
accepted identity. It rejects changed source, actor, task, agent,
result, controller, process, profile, allocation and receipt evidence. A
post-acceptance source change keeps the same intent pending. Execution
revalidation rejects removed or mismatched proof. The probe stops
deliberately after production identity initialization; it does not claim
provider execution. The final source passes the 75-case Retry group, the
focused two-case board/inherited-policy group and 19 canonical Retry
cases. Full workspace typecheck, build and map checks pass.
- Full workspace `pnpm -r typecheck` and `pnpm build` pass. Map,
module-boundary and diff checks pass. Full local `pnpm test:run` remains
unverified. The 256-case explicit-native suite cannot initialize its
packaged PostgreSQL because libzstd is absent; an existing
command-scoped fallback did not resolve it. Current-head remote
full-suite checks are required. Merged-source full GitHub+Retry (249),
separate Retry/HTTP (74), full wait/status (252), profile/history and
completion (75), and visible Retry hook/banner (12) checks pass. The
fresh owned PostgreSQL 18 databases applied all 377 migrations.
- Actual Rust Runner, OpenCode 1.18.34 and exact DeepSeek used unchanged
production-generated synthetic contexts. Resume finished with an
explicit server wait. The author-reply trial omitted continuation twice:
its first report failed the existing empty-criteria check; the corrected
second report still omitted continuation and received the exact server
wait. It completed once with no missing-wait loop or Done escalation.
Controller acknowledgments were synthetic; these probes do not establish
hosted publication or board state.
- Hosted combined testing proved automatic admission of the exact saved
author reply, a useful GitHub answer and committed passive wait. It
required two malformed-continuation rejections before a nullish fallback
succeeded. A later normal Pause/Resume completed in one finish with
explicit response_wake and no new review/comment. The admitted normal
Retry then exposed the execution guard gap repaired here. Those remain
checkpoint results. On the final combined source, renewed normal Retry
completed actual native provider work and GitHub publication, committed
the passive wait, released ownership and cleaned up its environment. A
subsequent ordinary unmentioned author reply woke once, made no GitHub
tool calls or new assessment, and completed its first finish with an
explicit response wake. Both exact leases expired with successful
cleanup and the queue was empty. However, final destination inspection
found a separate automatic fallback overwrote the existing summary with
a generic no-final-response message. External summary preservation
therefore remains failed acceptance; the separate publication repair in
#17726 owns that producer. Full combined publication preservation has
not passed. These hosted runs include the pinned-base native additions;
they do not qualify Copilot v63.
- Current-head automated review is 5/5 with zero unresolved threads. All
current-head CI gates passed (54 successful checks/contexts, four
skipped, none pending). The earlier login-cleanup idle-work assertion
also reproduced locally while its controlled baseline passed; its cause
remains unknown, but the exact assertion passed in the fresh
current-head CI shard. No timeout, assertion or production workaround
was added. This PR is ready for review within its source, authority,
native finalization, state, wake and cleanup scope. The independent
automatic fallback publication regression is tracked by #17726; no
complete combined workflow pass is claimed.

## Risks

- The latest saved review defines the durable head boundary. Ingress
must record a new head before admission. A reply receipt proves delivery
only for its exact saved source; a shared GitHub comment ID alone cannot
authorize this wait.
- Historical App recovery cannot infer identity without direct proof or
an earlier valid server response receipt. It never replays that
response.
- This does not infer review-score policy from instructions or override
Done. No deployment, merge or repair of historical tasks is claimed by
the PR.

## Model Used

OpenAI `gpt-6.1-sol`, with reasoning, repository inspection, code
editing and test execution. The context-window size is not exposed in
this session.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass (focused database and
native suites; full local suite limitation described above)
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green (verified on immutable latest
head)
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
(verified on the latest head)
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-10-26 22:22:47 -05:00

5119 lines
205 KiB
TypeScript

import { initializeRunIdentity } from "../run-identity.js";
import { createHash, randomUUID } from "node:crypto";
import { mkdir, mkdtemp, rm, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import path from "node:path";
import { and, eq, ne, sql } from "drizzle-orm";
import { createDatabaseResponseCapture, installDatabaseResponseCapture } from "@paperclipai/db/response-capture";
import { afterAll, beforeAll, describe, expect, it, vi } from "vitest";
import {
agentWakeupRequests,
agents,
authUsers,
approvals,
chatActions,
chatConversations,
chatDeliveries,
chatEndpointResources,
chatEndpoints,
chatExternalPrincipals,
chatGitHubConfigurations,
chatGitHubReviews,
chatIdentityLinks,
chatMessageLinks,
chatPublications,
companies,
companyMemberships,
completionContracts,
closeRegisteredClients,
createDb,
heartbeatRuns,
issueComments,
issueAttachments,
issueApprovals,
issueThreadInteractions,
issueQuestionResponseDeliveries,
issues,
nativeRunFinalizations,
nativeRunResults,
runIdentityContexts,
statusDecisions,
toolApplications,
toolConnections,
chatVoiceSessions, chatVoiceReplies, chatVoiceToolCalls,
workspaceOperations,
} from "@paperclipai/db";
import { validatePrpStructuredRunResult } from "../../vendor/paperclip-runner/index.js";
import type {
PrpStructuredRunResult,
PrpTerminalState,
} from "../../vendor/paperclip-runner/index.js";
import { startEmbeddedPostgresTestDatabase } from "../../__tests__/helpers/embedded-postgres.js";
import { finalizeNativeRun } from "./native-run-finalizer.js";
import { nativeCompletionFeedback } from "./native-completion-feedback.js";
import { hasPublishedGitHubReview, readInheritedGitHubReviewWait } from "../chat-channels/github/review-wait.js";
import * as chatWaitAuthority from "./chat-attachment-reuse.js";
import * as publishedReviewAuthority from "../chat-channels/github/review-wait.js";
import { toolAccessService } from "../tool-access.js";
import { PaperclipControlPlanePort } from "./paperclip-control-plane-port.js";
import {
authorizeNativeChatReviewPresentation,
hasMaterializedNativeReviewResponse,
} from "./native-chat-review-presentation.js";
import * as nativeChatReviewPresentation from "./native-chat-review-presentation.js";
import { resolveChatRunPresentationAuthorizationReason } from "../chat-run-publications.js";
import { resolveHeartbeatRunResponse } from "../heartbeat-run-summary.js";
import { issueService } from "../issues.js";
import { issueThreadInteractionService } from "../issue-thread-interactions.js";
import { reconcileNativeFinalizations } from "./native-finalization-reconciler.js";
import {
authorizeChatConversationForBoundRun,
isExternalChatWaitAuthorizationContention,
resolveExternalChatResponseWaitAuthorization,
} from "./chat-attachment-reuse.js";
import { attestReviewedExternalChatRun, buildPaperclipWakePayload } from "../heartbeat.js";
import { voiceCredentialFingerprint } from "../voice/voice-session-store.js";
import { nativeSha256 } from "./canonical.js";
import { questionResponseDeliveryValues } from "../question-response-delivery.js";
import { resolveExternalChatQuestionResponse } from "./external-chat-question-response.js";
import { listExternalReviewWaits } from "./external-review-wait.js";
import { materializeExternalChatQuestionResponseInput } from "./external-chat-question-response-input.js";
import * as nativeInteractionBridge from "./native-interaction-bridge.js";
import { defaultGitHubReviewPolicy, type AskUserQuestionsInteraction } from "@paperclipai/shared";
import { PaperclipRunnerToolAuthority } from "./paperclip-runner-tool-authority.js";
import { createLocalDiskStorageProvider } from "../../storage/local-disk-provider.js";
import { createStorageService } from "../../storage/service.js";
import { subscribeAllCompanyLiveEvents } from "../live-events.js";
describe("native external-chat response wait", () => {
const externalTestDatabaseUrl = process.env.PAPERCLIP_TEST_DATABASE_URL;
let temporary: Awaited<
ReturnType<typeof startEmbeddedPostgresTestDatabase>
> | null = null;
let db: ReturnType<typeof createDb>;
beforeAll(async () => {
if (externalTestDatabaseUrl) {
db = createDb(externalTestDatabaseUrl);
return;
}
temporary = await startEmbeddedPostgresTestDatabase(
"native-external-chat-wait-",
);
db = createDb(temporary.connectionString);
}, 30_000);
afterAll(async () => {
await temporary?.cleanup();
if (externalTestDatabaseUrl)
await closeRegisteredClients(externalTestDatabaseUrl);
});
async function seedWaitTurn(
provider:
| "telegram"
| "discord"
| "github"
| "microsoft-teams"
| "slack" = "telegram",
attentionRequests: PrpStructuredRunResult["attentionRequests"] = [],
resultOverrides: Partial<PrpStructuredRunResult> = {},
deferCompletion = false,
contractOverride?: { objective: string; criteria: Array<{ id: string; requirement: string }> },
) {
const companyId = randomUUID();
const agentId = randomUUID();
const issueId = randomUUID();
const runId = randomUUID();
const sessionId = randomUUID();
const runnerInstanceId = randomUUID();
const contractId = randomUUID();
const endpointId = randomUUID();
const resourceId = randomUUID();
const conversationId = randomUUID();
const principalId = randomUUID();
const commentId = randomUUID();
const deliveryId = randomUUID();
const userId = `wait-user-${randomUUID()}`;
const applicationId = randomUUID();
const connectionId = randomUUID();
const contractSha256 = `external-chat-wait-${randomUUID()}`;
// Keep prefix uniqueness as strong as the company's primary key. A five-
// digit random prefix can collide within this fixture-heavy suite.
const issuePrefix = `W${companyId.replaceAll("-", "").toUpperCase()}`;
await db.insert(companies).values({
id: companyId,
name: "External chat wait",
issuePrefix,
issueCounter: 1,
});
await db.insert(agents).values({
id: agentId,
companyId,
name: "Waiting chat agent",
adapterType: "paperclip_runner",
adapterConfig: { provider: "codex" },
runtimeConfig: {},
status: "active",
});
await db.insert(issues).values({
id: issueId,
companyId,
issueNumber: 1,
identifier: `${issuePrefix}-1`,
title: "Send the photo and wait",
status: "in_progress",
workMode: "standard",
assigneeAgentId: agentId,
});
await db.insert(completionContracts).values({
id: contractId,
companyId,
issueId,
revision: 1,
schemaVersion: "paperclip.completion-contract.v1",
policyVersion: "phase6-v3",
risk: "low",
completionAuthority: "agent_claim_policy",
incompleteCriteriaPolicy: "preserve_non_terminal",
contractJson: {
revision: "external-chat-wait-v1",
objective: "Send the requested photo and wait for the next message",
criteria: [
{ id: "response", requirement: "Return the requested photo" },
],
...contractOverride,
},
canonicalSha256: contractSha256,
createdByActorType: "system",
createdByActorId: "test",
});
await db.insert(heartbeatRuns).values({
id: runId,
companyId,
agentId,
status: "running",
runtimeMode: "native",
nativeIssueId: issueId,
nativeSessionId: sessionId,
runnerInstanceId,
completionContractId: contractId,
completionContractSha256: contractSha256,
contextSnapshot: {},
});
await db
.update(issues)
.set({ executionRunId: runId })
.where(eq(issues.id, issueId));
await db.insert(toolApplications).values({
id: applicationId,
companyId,
applicationKey: `chat:telegram:${endpointId}`,
name: "Telegram wait",
type: "chat",
status: "active",
});
await db.insert(toolConnections).values({
id: connectionId,
companyId,
applicationId,
name: "Telegram wait",
uid: `chat-telegram-${endpointId}`,
connectionPurpose: "channel",
transport: "chat_sdk",
status: "active",
enabled: true,
});
await db.insert(chatEndpoints).values({
id: endpointId,
companyId,
connectionId,
provider,
publicId: randomUUID(),
assignedAgentId: agentId,
status: "active",
providerAccountId: "telegram-bot",
allowDirectMessages: true,
allowUnlinkedPeople: false,
});
await db.insert(chatEndpointResources).values({
id: resourceId,
companyId,
endpointId,
type: provider === "github" ? "repository" : "direct_message",
providerResourceId:
provider === "github" ? "paperclip/test-repository" : "telegram-user",
label:
provider === "github"
? "Paperclip test repository"
: "Telegram direct message",
availability: "available",
enabled: true,
});
await db.insert(chatConversations).values({
id: conversationId,
companyId,
endpointId,
resourceId,
issueId,
externalConversationId: "telegram-user",
externalThreadId: "telegram:telegram-user",
sessionGeneration: 1,
externalLabel: "Telegram direct message",
isDirectMessage: provider !== "github",
state: "active",
});
await db.insert(chatExternalPrincipals).values({
id: principalId,
companyId,
provider,
providerAccountId: "telegram-bot",
externalId: "telegram-user",
kind: "user",
});
await db.insert(chatIdentityLinks).values({
companyId,
endpointId,
principalId,
paperclipUserId: userId,
status: "linked",
});
await db.insert(authUsers).values({ id: userId, name: "Review owner", email: `${userId}@example.test`,
createdAt: new Date(), updatedAt: new Date() });
await db.insert(companyMemberships).values({
companyId,
principalType: "user",
principalId: userId,
status: "active",
membershipRole: "member",
});
await db.insert(issueComments).values({
id: commentId,
companyId,
issueId,
authorType: "user",
authorUserId: userId,
body: "Send the photo, keep this task in progress, and wait.",
});
await db.insert(chatDeliveries).values({
id: deliveryId,
companyId,
endpointId,
conversationId,
principalId,
providerEventId: "telegram-wait-message",
deduplicationKey: "telegram-wait-message",
eventKind: "message",
normalizedEvent: {},
state: "processed",
attempts: 1,
processedAt: new Date(),
});
await db.insert(chatMessageLinks).values({
companyId,
endpointId,
conversationId,
deliveryId,
commentId,
providerMessageId: "telegram-user:1",
direction: "inbound",
});
await db
.update(heartbeatRuns)
.set({
contextSnapshot: {
source: `chat:${provider}`,
paperclipHarnessCheckedOut: true,
issueId,
wakeCommentId: commentId,
wakeCommentIds: [commentId],
paperclipWake: {
reason: "External chat message received",
externalChatProvider: provider,
checkedOutByHarness: true,
issue: { id: issueId, workMode: "standard" },
commentIds: [commentId],
},
},
})
.where(eq(heartbeatRuns.id, runId));
const port = new PaperclipControlPlanePort(db, {
companyId,
issueId,
runId,
agentId,
sessionId,
completionContractId: contractId,
completionContractSha256: contractSha256,
sourceInstanceId: runnerInstanceId,
controlPlaneSourceInstanceId: `wait-control-${runId}`,
});
await port.openRun({
identity: { companyId, issueId, runId, agentId, sessionId },
backendKind: "mock",
sourceInstanceId: runnerInstanceId,
});
const result: PrpStructuredRunResult = {
schema: "paperclip.run_result.v1",
reportedWorkDisposition: "yielded",
summary:
"The requested photo is prepared. I will wait for your next message.",
completionClaim: {
contractRevision: "external-chat-wait-v1",
objectiveSatisfied: true,
criteria: [
{
criterionId: "response",
status: "satisfied",
evidenceRefs: [],
},
],
remainingWork: [],
},
evidence: [],
verification: [],
attentionRequests,
artifacts: [],
continuation: {
kind: "response_wake",
summary: "Wait for the next authorized Telegram message.",
idempotencyKey: `telegram-response-wait:${conversationId}`,
},
...resultOverrides,
};
const terminal: PrpTerminalState = {
schema: "paperclip.prp.terminal.v1",
turnTerminalState: "completed",
runTerminalState: "succeeded",
reportedWorkDisposition: result.reportedWorkDisposition,
workAssessmentId: randomUUID(),
statusDecisionId: randomUUID(),
};
if (!deferCompletion) await port.completeRun({
result,
terminal,
callerResultId: `wait-result-${runId}`,
});
await db
.update(heartbeatRuns)
.set({
status: "running",
resultJson: {
nativeResult: result as unknown as Record<string, unknown>,
},
updatedAt: new Date(),
})
.where(eq(heartbeatRuns.id, runId));
return {
port, result, terminal,
agentId,
companyId,
conversationId,
endpointId,
deliveryId,
issueId,
resourceId,
runId,
userId,
commentId,
principalId,
};
}
async function placeWaitTurnInSetupTest(
fixture: Awaited<ReturnType<typeof seedWaitTurn>>,
options: {
credentialFingerprint?: string;
generation?: number;
processedAt?: Date;
receivedAt?: Date;
setupStep?: "provider_setup" | "test";
testStartedAt?: Date;
} = {},
) {
const testStartedAt =
options.testStartedAt ?? new Date(Date.now() - 5_000);
const receivedAt = options.receivedAt ?? new Date(Date.now() - 4_000);
const processedAt = options.processedAt ?? new Date(Date.now() - 3_000);
const generation = options.generation ?? 1;
await db
.update(chatEndpoints)
.set({
status: "verifying",
setup: {
step: options.setupStep ?? "test",
testStartedAt: testStartedAt.toISOString(),
runtimeGeneration: generation,
} as (typeof chatEndpoints.$inferSelect)["setup"] & {
runtimeGeneration: number;
},
updatedAt: new Date(),
})
.where(eq(chatEndpoints.id, fixture.endpointId));
await db
.update(chatDeliveries)
.set({
normalizedEvent: {
runtimeContext: {
generation,
credentialFingerprint:
options.credentialFingerprint ?? "a".repeat(64),
},
},
processedAt,
receivedAt,
updatedAt: new Date(),
})
.where(eq(chatDeliveries.id, fixture.deliveryId));
return { processedAt, receivedAt, testStartedAt };
}
async function seedAnsweredChatTurn(
provider: "telegram" | "discord" | "slack" = "telegram",
target?: {
fixture: Awaited<ReturnType<typeof seedWaitTurn>>;
gate: Awaited<ReturnType<typeof seedPriorCompletionReview>>;
},
responseKind: "button" | "form" | "multi" = "button",
) {
const fixture = target?.fixture ?? await seedWaitTurn(provider);
const gate = target?.gate ?? await seedPriorCompletionReview(fixture);
const [current] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const sourceRunId = randomUUID();
const interactionId = randomUUID();
const wakeId = randomUUID();
const publicationId = randomUUID();
const actionId = randomUUID();
const formActionId = `pcfs:${"A".repeat(22)}`;
const selectFieldId = `pcff:${"B".repeat(22)}`;
const textFieldId = `pcff:${"C".repeat(22)}`;
const selectValue = `pcfo:${"D".repeat(22)}`;
const formExpiresAt = new Date(Date.now() + 60_000).toISOString();
const interactionPayload =
responseKind === "form"
? {
version: 1 as const,
questions: [
{
id: "environment",
prompt: "Choose an environment",
selectionMode: "single" as const,
required: true,
allowOther: false,
options: [{ id: "cedar", label: "Cedar" }],
},
{
id: "note",
prompt: "Enter a release note",
selectionMode: "single" as const,
required: true,
allowOther: true,
options: [
{
id: "__paperclip_text__",
label: "Type an answer",
freeText: true,
},
],
},
],
}
: {
version: 1 as const,
questions: [
{
id: "color",
prompt: "Choose a color",
selectionMode: responseKind === "multi" ? "multi" as const : "single" as const,
required: true,
allowOther: false,
options: [
{ id: "cobalt", label: "Cobalt" },
{ id: "amber", label: "Amber" },
],
},
],
};
const interactionResult =
responseKind === "form"
? {
version: 1 as const,
answers: [
{ questionId: "environment", optionIds: ["cedar"] },
{
questionId: "note",
optionIds: [],
otherText: "cobalt lantern82",
},
],
}
: {
version: 1 as const,
answers: [{ questionId: "color", optionIds: responseKind === "multi" ? ["cobalt", "amber"] : ["cobalt"] }],
};
await db.insert(heartbeatRuns).values({
id: sourceRunId,
companyId: fixture.companyId,
agentId: fixture.agentId,
status: "succeeded",
runtimeMode: "native",
nativeIssueId: fixture.issueId,
contextSnapshot: current!.contextSnapshot,
});
const [interaction] = await db
.insert(issueThreadInteractions)
.values({
id: interactionId,
companyId: fixture.companyId,
issueId: fixture.issueId,
kind: "ask_user_questions",
status: "answered",
sourceRunId,
createdByAgentId: fixture.agentId,
resolvedByUserId: fixture.userId,
resolvedAt: new Date(),
idempotencyKey: `color-${interactionId}`,
payload: interactionPayload,
result: interactionResult,
})
.returning();
const [responseDelivery] = await db
.insert(issueQuestionResponseDeliveries)
.values({
...questionResponseDeliveryValues(
interaction! as unknown as AskUserQuestionsInteraction,
),
status: "fallback_queued",
deliveryMode: "wake_fallback",
targetRunId: fixture.runId,
attemptCount: 1,
acknowledgedAt: new Date(),
})
.returning();
await db.insert(chatPublications).values({
id: publicationId,
companyId: fixture.companyId,
endpointId: fixture.endpointId,
conversationId: fixture.conversationId,
issueId: fixture.issueId,
state: "published",
idempotencyKey: `card-${interactionId}`,
providerMessageId: "question-card",
publishedAt: new Date(),
payload: { interactionId } as never,
});
await db.insert(chatActions).values({
id: actionId,
companyId: fixture.companyId,
endpointId: fixture.endpointId,
conversationId: fixture.conversationId,
principalId: fixture.principalId,
kind:
responseKind === "form" ? "question_form_submit" : "question_answer",
status: "processed",
providerActionId:
responseKind === "form" ? formActionId : `answer-${interactionId}`,
payload:
responseKind === "form"
? {
version: 1,
publicationId,
interactionId,
formActionId,
expiresAt: formExpiresAt,
fields: [
{
fieldId: selectFieldId,
kind: "single_select",
questionId: "environment",
required: true,
options: [{ optionId: "cedar", value: selectValue }],
},
{
fieldId: textFieldId,
kind: "text",
questionId: "note",
required: true,
minLength: 0,
maxLength: 3_000,
inputType: "text",
},
],
}
: {
version: 1,
interactionId,
publicationId,
questionId: "color",
...(responseKind === "multi" ? { selectionOptions: [{ value: "opaque-cobalt", optionId: "cobalt" }, { value: "opaque-amber", optionId: "amber" }], selectionValues: ["opaque-cobalt", "opaque-amber"] } : { optionId: "cobalt" }),
},
result:
responseKind === "form"
? { code: "question_form_answered", interactionId }
: { interactionId, interactionStatus: "answered" },
});
await db.insert(agentWakeupRequests).values({
id: wakeId,
companyId: fixture.companyId,
agentId: fixture.agentId,
source: "automation",
triggerDetail: "system",
reason: "issue_commented",
requestedByActorType: "user",
requestedByActorId: fixture.userId,
idempotencyKey: `question-response:${interactionId}`,
status: "claimed",
runId: fixture.runId,
payload: {
issueId: fixture.issueId,
interactionId,
sourceRunId,
sourceCommentId: fixture.commentId,
mutation: "interaction",
externalChatContinuation: true,
},
});
const context: Record<string, unknown> = {
issueId: fixture.issueId,
taskId: fixture.issueId,
source: "issue.interaction.respond",
wakeReason: "issue_commented",
interactionId,
interactionKind: "ask_user_questions",
interactionStatus: "answered",
sourceRunId,
sourceCommentId: fixture.commentId,
wakeCommentId: fixture.commentId,
wakeCommentIds: [fixture.commentId],
externalChatContinuation: true,
};
await db
.update(heartbeatRuns)
.set({ wakeupRequestId: wakeId, contextSnapshot: context, status: "running" })
.where(eq(heartbeatRuns.id, fixture.runId));
return {
...fixture,
gate,
context,
interactionId,
sourceRunId,
responseDeliveryId: responseDelivery!.id,
wakeId,
publicationId,
actionId,
};
}
async function seedBoardAnsweredTurn(withLink = true, provider: "github" | "slack" = "github", oversized = false) {
const fixture = await seedWaitTurn(provider);
if (provider === "slack") {
await db.update(chatEndpointResources).set({ type: "channel" }).where(eq(chatEndpointResources.id, fixture.resourceId));
await db.update(chatConversations).set({ isDirectMessage: false }).where(eq(chatConversations.id, fixture.conversationId));
}
const gate = await seedPriorCompletionReview(fixture);
const [current] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const sourceRunId = randomUUID();
await db.insert(heartbeatRuns).values({
id: sourceRunId,
companyId: fixture.companyId,
agentId: fixture.agentId,
status: "succeeded",
runtimeMode: "native",
nativeIssueId: fixture.issueId,
contextSnapshot: current!.contextSnapshot,
});
await db
.update(chatEndpoints)
.set({
setup: {
step: "complete",
runtimeGeneration: 1,
} as typeof chatEndpoints.$inferInsert.setup,
})
.where(eq(chatEndpoints.id, fixture.endpointId));
await db
.update(chatDeliveries)
.set({
normalizedEvent: {
runtimeContext: {
generation: 1,
credentialFingerprint: "a".repeat(64),
},
},
})
.where(eq(chatDeliveries.id, fixture.deliveryId));
const interactionSvc = issueThreadInteractionService(db);
const previousPublicUrl = process.env.PAPERCLIP_PUBLIC_URL;
process.env.PAPERCLIP_PUBLIC_URL = withLink
? "https://board.paperclip.example"
: "http://127.0.0.1:3103";
let interaction: Awaited<ReturnType<typeof interactionSvc.create>>;
try {
interaction = await interactionSvc.create(
{ id: fixture.issueId, companyId: fixture.companyId },
{
kind: "ask_user_questions",
sourceRunId,
continuationPolicy: "wake_assignee",
payload: {
version: 1,
questions: [
{
id: "color",
prompt: "Choose a color",
// Multi-select plus a custom answer intentionally needs Board.
selectionMode: provider === "slack" ? "multi" : "single",
required: true,
allowOther: provider === "slack" && !oversized,
options: [
{ id: "cobalt", label: "Cobalt" },
{ id: "amber", label: "Amber" },
...(oversized ? Array.from({ length: 9 }, (_, index) => ({
id: `option-${index}`, label: `Option ${index}`, description: "Detail ".repeat(550),
})) : []),
],
},
],
},
},
{ agentId: fixture.agentId, runId: sourceRunId },
);
} finally {
if (previousPublicUrl === undefined)
delete process.env.PAPERCLIP_PUBLIC_URL;
else process.env.PAPERCLIP_PUBLIC_URL = previousPublicUrl;
}
const publicationKey = `interaction:${interaction.id}:${fixture.endpointId}`;
const [publication] = await db
.update(chatPublications)
.set({
state: "published",
providerMessageId: `${provider}-question`,
publishedAt: new Date(),
})
.where(eq(chatPublications.idempotencyKey, publicationKey))
.returning();
expect(publication).toBeTruthy();
await interactionSvc.answerQuestions(
{ id: fixture.issueId, companyId: fixture.companyId },
interaction.id,
{ answers: [{ questionId: "color", optionIds: ["cobalt"] }] },
{ userId: fixture.userId },
);
const [delivery] = await db
.update(issueQuestionResponseDeliveries)
.set({
status: "fallback_queued",
deliveryMode: "wake_fallback",
targetRunId: fixture.runId,
attemptCount: 1,
acknowledgedAt: new Date(),
})
.where(eq(issueQuestionResponseDeliveries.interactionId, interaction.id))
.returning();
const wakeId = randomUUID();
await db.insert(agentWakeupRequests).values({
id: wakeId,
companyId: fixture.companyId,
agentId: fixture.agentId,
source: "automation",
triggerDetail: "system",
reason: "issue_commented",
requestedByActorType: "user",
requestedByActorId: fixture.userId,
idempotencyKey: `question-response:${interaction.id}`,
status: "claimed",
runId: fixture.runId,
payload: {
issueId: fixture.issueId,
interactionId: interaction.id,
sourceRunId,
sourceCommentId: fixture.commentId,
mutation: "interaction",
externalChatContinuation: true,
},
});
const context: Record<string, unknown> = {
issueId: fixture.issueId,
taskId: fixture.issueId,
source: "issue.interaction.respond",
wakeReason: "issue_commented",
interactionId: interaction.id,
interactionKind: "ask_user_questions",
interactionStatus: "answered",
sourceRunId,
sourceCommentId: fixture.commentId,
wakeCommentId: fixture.commentId,
wakeCommentIds: [fixture.commentId],
externalChatContinuation: true,
};
await db
.update(heartbeatRuns)
.set({
wakeupRequestId: wakeId,
contextSnapshot: context,
status: "running",
})
.where(eq(heartbeatRuns.id, fixture.runId));
return {
...fixture,
gate,
context,
sourceRunId,
interactionId: interaction.id,
publicationId: publication!.id,
responseDeliveryId: delivery!.id,
wakeId,
};
}
it.each(["telegram", "github", "github_board_answer", "slack_board_answer"] as const)(
"reads small current external review waits without saved prompts (%s)", async source => {
const fixture = await (async () => {
if (source === "github_board_answer" || source === "slack_board_answer") {
const answered = await seedBoardAnsweredTurn(true, source === "github_board_answer" ? "github" : "slack");
await attestAnswer(answered);
return answered;
}
return seedWaitTurn(source);
})();
const finalized = await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect(finalized.phase).toBe("committed");
await db.update(statusDecisions).set({ toStatus: "in_review" }).where(eq(statusDecisions.id, finalized.decisionId!));
await db.update(issues).set({ status: "in_review" }).where(eq(issues.id, fixture.issueId));
const [decision] = await db.select({ reason: statusDecisions.reasonCode, state: statusDecisions.applicationState })
.from(statusDecisions).where(eq(statusDecisions.id, finalized.decisionId!));
const governed = source === "github_board_answer" || source === "slack_board_answer";
expect(decision).toEqual({ reason: governed ? "governed_response_waiting" : "external_chat_response_waiting", state: "applied" });
const [run] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
expect(run!.status).toBe("succeeded");
const contextSnapshot = { ...run!.contextSnapshot, savedPrompt: "external-wait-saved-input🙂".repeat(30_000) };
await db.update(heartbeatRuns).set({ contextSnapshot }).where(eq(heartbeatRuns.id, fixture.runId));
await authorizeChatConversationForBoundRun(db, fixture, contextSnapshot, "read");
const observed = createDb(temporary?.connectionString ?? externalTestDatabaseUrl!);
const events: Array<Record<string, unknown>> = [], now = Date.now();
const capture = createDatabaseResponseCapture({ start: now - 100, end: now + 59_000 }, event => events.push(event));
installDatabaseResponseCapture(observed, { issues, heartbeatRuns }, capture);
try {
expect(await listExternalReviewWaits(observed, fixture.companyId, [fixture.issueId])).toEqual(governed ? [] : [
expect.objectContaining({ companyId: fixture.companyId, issueId: fixture.issueId }),
]);
capture.flush();
const reads = events.filter(event => event.table === "issues" && event.operation === "select");
expect(reads.length).toBeGreaterThan(0);
expect(reads.reduce((bytes, event) => bytes + Number(event.responseValueBytes), 0)).toBeLessThan(4096);
expect(await listExternalReviewWaits(observed, randomUUID(), [fixture.issueId])).toEqual([]);
expect(await listExternalReviewWaits(observed, fixture.companyId, [])).toEqual([]);
await db.update(companyMemberships).set({ status: "suspended" }).where(eq(companyMemberships.companyId, fixture.companyId));
expect(await listExternalReviewWaits(observed, fixture.companyId, [fixture.issueId])).toEqual([]);
expect((await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId)))[0]?.contextSnapshot).toEqual(contextSnapshot);
} finally { capture.flush(); await observed.$client.end(); }
},
);
it("resumes an oversized Slack multi-select answered through its Paperclip handoff", async () => {
const fixture = await seedBoardAnsweredTurn(true, "slack", true);
const [publication] = await db.select().from(chatPublications).where(eq(chatPublications.id, fixture.publicationId));
expect(publication!.payload.card?.actions).toEqual([
{ type: "link", label: "Open in Paperclip", url: expect.stringContaining(fixture.issueId) },
]);
expect(publication!.payload.card?.selection).toBeUndefined();
await expect(db.select().from(chatActions).where(eq(chatActions.companyId, fixture.companyId))).resolves.toEqual([]);
await attestAnswer(fixture);
const responses = await materializeExternalChatQuestionResponseInput({ db, binding: fixture, contextSnapshot: fixture.context });
expect(responses).toHaveLength(1);
expect(JSON.stringify(responses)).toContain("cobalt");
});
it.each((["github", "slack"] as const).flatMap(provider => [true, false].map(withLink => ({ provider, withLink }))))(
"attests a native $provider question answered in Board without inventing a provider action (link: $withLink)",
async ({ provider, withLink }) => {
const fixture = await seedBoardAnsweredTurn(withLink, provider);
await expect(
db
.select()
.from(chatActions)
.where(eq(chatActions.companyId, fixture.companyId)),
).resolves.toEqual([]);
expect(
await resolveExternalChatQuestionResponse(
db,
fixture,
fixture.context,
"read",
true,
),
).not.toBeNull();
await attestAnswer(fixture);
expect(fixture.context.paperclipWake).toMatchObject({
externalChatProvider: provider,
});
await db
.update(heartbeatRuns)
.set({ contextSnapshot: fixture.context })
.where(eq(heartbeatRuns.id, fixture.runId));
const responses = await materializeExternalChatQuestionResponseInput({
db,
binding: fixture,
contextSnapshot: fixture.context,
});
expect(responses).toHaveLength(1);
expect(JSON.stringify(responses)).toContain("cobalt");
expect(responses[0]!.interactionId).toBe(fixture.interactionId);
const resultJson = await finishReviewResponse(fixture);
expect(resultJson.externalChatReviewPresentation).toMatchObject({
gateId: fixture.gate.id,
});
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
const publications = await db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId));
await finishReviewResponse(fixture);
await expect(
db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId)),
).resolves.toEqual(publications);
await expect(
db
.select({ status: issueThreadInteractions.status })
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.gate.id)),
).resolves.toEqual([{ status: "pending" }]);
},
);
it.each((["github", "slack"] as const).flatMap(provider => ([
"different_user",
"viewer",
"membership_revoked",
"identity_revoked",
"generation",
"missing_runtime",
"wrong_target",
"wrong_wake",
"unpublished",
"provider_action",
"deleted_source",
"endpoint_paused",
"resource_revoked",
"answer_hash",
] as const).filter(change => provider === "github" || (change !== "generation" && change !== "missing_runtime")).map(change => ({ provider, change }))))(
"rejects $provider Board answer authority after $change",
async ({ provider, change }) => {
const fixture = await seedBoardAnsweredTurn(true, provider);
if (change === "different_user") {
const otherUser = `other-${randomUUID()}`;
await db
.insert(companyMemberships)
.values({
companyId: fixture.companyId,
principalType: "user",
principalId: otherUser,
status: "active",
membershipRole: "member",
});
await db
.update(issueThreadInteractions)
.set({ resolvedByUserId: otherUser })
.where(eq(issueThreadInteractions.id, fixture.interactionId));
await db
.update(agentWakeupRequests)
.set({ requestedByActorId: otherUser })
.where(eq(agentWakeupRequests.id, fixture.wakeId));
const [interaction] = await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.interactionId));
const receipt = questionResponseDeliveryValues(
interaction! as unknown as AskUserQuestionsInteraction,
);
await db
.update(issueQuestionResponseDeliveries)
.set({ payloadSha256: receipt.payloadSha256 })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
}
if (change === "viewer" || change === "membership_revoked")
await db
.update(companyMemberships)
.set(
change === "viewer"
? { membershipRole: "viewer" }
: { status: "suspended" },
)
.where(eq(companyMemberships.companyId, fixture.companyId));
if (change === "identity_revoked")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.endpointId, fixture.endpointId));
if (change === "generation")
await db
.update(chatEndpoints)
.set({
setup: {
step: "complete",
runtimeGeneration: 2,
} as typeof chatEndpoints.$inferInsert.setup,
})
.where(eq(chatEndpoints.id, fixture.endpointId));
if (change === "missing_runtime")
await db
.update(chatDeliveries)
.set({ normalizedEvent: {} })
.where(eq(chatDeliveries.id, fixture.deliveryId));
if (change === "wrong_target")
await db
.update(issueQuestionResponseDeliveries)
.set({ targetRunId: fixture.sourceRunId })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
if (change === "wrong_wake")
await db
.update(agentWakeupRequests)
.set({ runId: fixture.sourceRunId })
.where(eq(agentWakeupRequests.id, fixture.wakeId));
if (change === "unpublished")
await db
.update(chatPublications)
.set({ state: "pending", publishedAt: null, providerMessageId: null })
.where(eq(chatPublications.id, fixture.publicationId));
if (change === "provider_action")
await db
.insert(chatActions)
.values({
companyId: fixture.companyId,
endpointId: fixture.endpointId,
conversationId: fixture.conversationId,
principalId: fixture.principalId,
kind: "question_answer",
status: "processed",
providerActionId: randomUUID(),
payload: {
version: 1,
interactionId: fixture.interactionId,
publicationId: fixture.publicationId,
},
result: {},
});
if (change === "deleted_source")
await db
.update(issueComments)
.set({ deletedAt: new Date() })
.where(eq(issueComments.id, fixture.commentId));
if (change === "endpoint_paused")
await db
.update(chatEndpoints)
.set({ status: "paused" })
.where(eq(chatEndpoints.id, fixture.endpointId));
if (change === "resource_revoked")
await db
.update(chatEndpointResources)
.set({ enabled: false })
.where(eq(chatEndpointResources.id, fixture.resourceId));
if (change === "answer_hash")
await db
.update(issueQuestionResponseDeliveries)
.set({ payloadSha256: "b".repeat(64) })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).toBe(false);
expect(
fixture.context.paperclipExternalChatQuestionResponse,
).toBeUndefined();
await expect(
db
.select({ status: issueThreadInteractions.status })
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.gate.id)),
).resolves.toEqual([{ status: "pending" }]);
},
);
it.each((["github", "slack"] as const).flatMap(provider => (["generation", "identity", "membership", "reach"] as const).filter(change => provider === "github" || change !== "generation").map(change => ({ provider, change }))))(
"rechecks $provider Board answer $change at publication after native completion",
async ({ provider, change }) => {
const fixture = await seedBoardAnsweredTurn(true, provider);
await attestAnswer(fixture);
const resultJson = await finishReviewResponse(fixture);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
if (change === "generation")
await db
.update(chatEndpoints)
.set({
setup: {
step: "complete",
runtimeGeneration: 2,
} as typeof chatEndpoints.$inferInsert.setup,
})
.where(eq(chatEndpoints.id, fixture.endpointId));
if (change === "identity")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.endpointId, fixture.endpointId));
if (change === "membership")
await db
.update(companyMemberships)
.set({ membershipRole: "viewer" })
.where(eq(companyMemberships.companyId, fixture.companyId));
if (change === "reach")
await db
.update(chatEndpointResources)
.set({ enabled: false })
.where(eq(chatEndpointResources.id, fixture.resourceId));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(false);
},
);
it.each((["native", "legacy"] as const).flatMap(sourceRuntime => (["voice", "board"] as const).map(answerRoute => ({sourceRuntime, answerRoute}))))("authorizes $sourceRuntime Speko question continuation from its exact $answerRoute answer", async ({sourceRuntime, answerRoute}) => {
const f = await seedAnsweredChatTurn();
const [endpoint] = await db.select().from(chatEndpoints).where(eq(chatEndpoints.id, f.endpointId));
const [connection] = await db.update(toolConnections).set({ transport: "voice", status: "active", enabled: true }).where(eq(toolConnections.id, endpoint.connectionId)).returning();
await db.update(chatEndpoints).set({ provider: "speko", setup: { step: "complete", runtimeGeneration: 1 } as typeof endpoint.setup }).where(eq(chatEndpoints.id, f.endpointId));
const [source] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, f.sourceRunId));
const sourceContext = JSON.parse(JSON.stringify(source.contextSnapshot).replaceAll("chat:telegram", "chat:speko").replaceAll('"telegram"', '"speko"'));
await db.update(heartbeatRuns).set({ contextSnapshot: sourceContext, runtimeMode: sourceRuntime, nativeIssueId: sourceRuntime === "native" ? f.issueId : null }).where(eq(heartbeatRuns.id, f.sourceRunId));
const resolve = () => resolveExternalChatQuestionResponse(db, f, f.context, "read", true);
expect(await resolve()).toBeNull();
const [interaction] = await db.select().from(issueThreadInteractions).where(eq(issueThreadInteractions.id, f.interactionId));
const sessionId = randomUUID();
await db.insert(chatVoiceSessions).values({ id: sessionId, companyId: f.companyId, endpointId: f.endpointId,
conversationId: f.conversationId, issueId: f.issueId, assignedAgentId: f.agentId, callerId: f.userId, callerAuthority: "member",
mode: "browser", state: "ended", providerSessionId: randomUUID(), generation: 1, credentialFingerprint: voiceCredentialFingerprint(connection.credentialSecretRefs),
toolTokenHash: "a".repeat(64), idempotencyKey: randomUUID(), requestFingerprint: "fixture", expiresAt: new Date(Date.now() + 60000) });
await db.insert(chatVoiceReplies).values({ companyId: f.companyId, sessionId, publicationId: f.publicationId, cursor: 1,
deliveredAt: new Date(interaction.resolvedAt!.getTime() - 1) });
const [receipt] = answerRoute === "voice" ? await db.insert(chatVoiceToolCalls).values({ companyId: f.companyId, sessionId, providerToolCallId: randomUUID(),
webhookId: randomUUID(), fingerprint: "signed-tool-fingerprint", tool: "answer_question",
response: { status: "answered", interactionId: f.interactionId, resultSha256: nativeSha256(interaction.result) } }).returning() : [];
if (answerRoute === "board") {
await db.update(chatExternalPrincipals).set({ provider: "speko", externalId: `voice:${sessionId}` }).where(eq(chatExternalPrincipals.id, f.principalId));
await db.insert(chatVoiceToolCalls).values({companyId: f.companyId, sessionId, providerToolCallId: randomUUID(),
webhookId: randomUUID(), fingerprint: "accepted-source-request", tool: "submit_request", response: {status: "accepted"}});
}
// The answer also publishes a terminal acknowledgment. It cannot compete
// with the pre-answer presentation as authority for the continuation.
const [terminal] = await db.insert(chatPublications).values({ companyId: f.companyId, endpointId: f.endpointId,
conversationId: f.conversationId, issueId: f.issueId, idempotencyKey: randomUUID(), state: "published",
payload: { text: "Answered: High.", interactionId: f.interactionId } }).returning();
await db.insert(chatVoiceReplies).values({ companyId: f.companyId, sessionId, publicationId: terminal.id, cursor: 2 });
expect(await resolve()).toMatchObject({ provider: "speko", marker: { interactionId: f.interactionId } });
// Voice authority is the authenticated session, not a synthetic provider
// identity link. Exercise the real native run attestation and recheck path.
await db.update(chatExternalPrincipals).set({ provider: "speko", externalId: `voice:${sessionId}` }).where(eq(chatExternalPrincipals.id, f.principalId));
await db.delete(chatIdentityLinks).where(eq(chatIdentityLinks.principalId, f.principalId));
await attestAnswer(f);
expect(await authorizeChatConversationForBoundRun(db, f, f.context, "read")).toMatchObject({ conversationId: f.conversationId });
await db.update(companyMemberships).set({ status: "removed" }).where(eq(companyMemberships.principalId, f.userId));
await expect(authorizeChatConversationForBoundRun(db, f, f.context, "read")).rejects.toThrow(answerRoute === "board" ? "paperclip_runner_chat_attachment_binding_denied" : "paperclip_runner_chat_attachment_principal_denied");
await db.update(companyMemberships).set({ status: "active" }).where(eq(companyMemberships.principalId, f.userId));
if (receipt) await db.update(chatVoiceToolCalls).set({ response: { ...receipt.response, resultSha256: "different-answer" } }).where(eq(chatVoiceToolCalls.id, receipt.id));
else await db.update(issueComments).set({authorUserId: randomUUID()}).where(eq(issueComments.id, f.commentId));
expect(await resolve()).toBeNull();
});
it.each(["valid", "wrong_task", "revoked_identity", "wrong_answer_actor"] as const)(
"attests legacy Slack question continuations only with the durable source binding: %s",
async (condition) => {
const fixture = await seedAnsweredChatTurn("slack");
const [source] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.sourceRunId));
const context = source!.contextSnapshot as Record<string, unknown>;
await db.update(heartbeatRuns).set({
runtimeMode: "legacy", nativeIssueId: null,
contextSnapshot: { ...context, paperclipWake: {
...(context.paperclipWake as Record<string, unknown>),
issue: { id: condition === "wrong_task" ? randomUUID() : fixture.issueId },
} },
}).where(eq(heartbeatRuns.id, fixture.sourceRunId));
if (condition === "revoked_identity") {
await db.update(chatIdentityLinks).set({ status: "revoked" }).where(eq(chatIdentityLinks.principalId, fixture.principalId));
}
if (condition === "wrong_answer_actor") {
await db.update(agentWakeupRequests).set({ requestedByActorId: "another-user" }).where(eq(agentWakeupRequests.id, fixture.wakeId));
}
expect(await attestReviewedExternalChatRun({ db, ...fixture, contextSnapshot: fixture.context })).toBe(condition === "valid");
if (condition === "valid") {
expect(fixture.context.paperclipExternalChatQuestionResponse).toMatchObject({ interactionId: fixture.interactionId });
} else {
expect(fixture.context.paperclipExternalChatQuestionResponse).toBeUndefined();
}
},
);
async function attestAnswer(
fixture:
| Awaited<ReturnType<typeof seedAnsweredChatTurn>>
| Awaited<ReturnType<typeof seedBoardAnsweredTurn>>,
) {
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).toBe(true);
expect(fixture.context.paperclipExternalChatQuestionResponse).toMatchObject(
{
schema: "paperclip.external_chat_question_response.v1",
interactionId: fixture.interactionId,
sourceRunId: fixture.sourceRunId,
responseDeliveryId: fixture.responseDeliveryId,
},
);
fixture.context.paperclipExternalChatExecutionBound = true;
fixture.context.paperclipWake = await buildPaperclipWakePayload({
db,
companyId: fixture.companyId,
agentId: fixture.agentId,
runId: fixture.runId,
contextSnapshot: fixture.context,
});
await db
.update(heartbeatRuns)
.set({ contextSnapshot: fixture.context })
.where(eq(heartbeatRuns.id, fixture.runId));
}
async function withAnsweredFileTool(
provider: "telegram" | "discord" | "github",
tool: "register_deliverable" | "reuse_chat_attachment",
check: (input: {
fixture:
| Awaited<ReturnType<typeof seedAnsweredChatTurn>>
| Awaited<ReturnType<typeof seedBoardAnsweredTurn>>;
invoke: () => Promise<unknown>;
}) => Promise<void>,
) {
const fixture =
provider === "github"
? await seedBoardAnsweredTurn()
: await seedAnsweredChatTurn(provider);
await attestAnswer(fixture);
const root = await mkdtemp(path.join(tmpdir(), "answered-chat-file-"));
try {
const workspaceRoot = path.join(root, "workspace");
await mkdir(workspaceRoot);
const storage = createStorageService(
createLocalDiskStorageProvider(path.join(root, "storage")),
);
const body = Buffer.from("Cobalt\n");
const filename = "answer.txt";
await writeFile(path.join(workspaceRoot, filename), body);
const runner = new PaperclipRunnerToolAuthority(db, {
...fixture,
workspaceRoot,
storage,
});
let sourceAttachmentId: string | null = null;
if (tool === "reuse_chat_attachment") {
const stored = await storage.putFile({
companyId: fixture.companyId,
namespace: `issues/${fixture.issueId}`,
originalFilename: filename,
contentType: "text/plain",
body,
});
const attachment = await issueService(db).createAttachment({
issueId: fixture.issueId,
issueCommentId: fixture.commentId,
provider: stored.provider,
objectKey: stored.objectKey,
contentType: stored.contentType,
byteSize: stored.byteSize,
sha256: stored.sha256,
originalFilename: stored.originalFilename,
createdByUserId: fixture.userId,
});
sourceAttachmentId = attachment.id;
}
await check({
fixture,
invoke: () =>
runner.execute({
tool,
callId: randomUUID(),
arguments: {
idempotencyKey: "answer-file-v1",
title: "Chosen color",
...(tool === "register_deliverable"
? {
filename,
contentRef: filename,
contentType: "text/plain",
byteSize: body.length,
sha256: createHash("sha256").update(body).digest("hex"),
}
: {
sourceCommentId: fixture.commentId,
attachmentId: sourceAttachmentId,
}),
},
}),
});
} finally {
await rm(root, { recursive: true, force: true });
}
}
it.each(
(["direct", "answered_question"] as const).flatMap((source) =>
(["error", "paused", "terminated", "pending_approval"] as const).map(
(status) => ({ source, status }),
),
),
)(
"uses current invokability for pre-start reviewed GitHub $source with agent $status",
async ({ source, status }) => {
const fixture =
source === "direct"
? await seedWaitTurn("github")
: await seedBoardAnsweredTurn();
if (source === "direct") await seedPriorCompletionReview(fixture);
// A previous run's failure projects `error` onto the agent. The next
// already-claimed turn attests before execution-start changes it to
// `running`; this does not grant permission to clear an operator gate.
await db
.update(agents)
.set({
status,
errorReason: status === "error" ? "Prior run failed" : null,
})
.where(eq(agents.id, fixture.agentId));
const [run] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const [issue] = await db
.select()
.from(issues)
.where(eq(issues.id, fixture.issueId));
const gates = await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.issueId, fixture.issueId));
expect(issue).toMatchObject({
status: "in_review",
assigneeAgentId: fixture.agentId,
executionRunId: fixture.runId,
});
expect(gates.some((gate) => gate.status === "pending")).toBe(true);
const context = structuredClone(run!.contextSnapshot!);
const onQuestionResponseAttested = vi.fn();
await expect(
attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: context,
onQuestionResponseAttested,
}),
).resolves.toBe(status === "error");
expect(onQuestionResponseAttested).toHaveBeenCalledTimes(
status === "error" && source === "answered_question" ? 1 : 0,
);
// Attestation proves the current exact binding without checking out,
// approving, retiring, or otherwise rewriting any durable authority.
expect(
await db.select().from(issues).where(eq(issues.id, fixture.issueId)),
).toEqual([issue]);
expect(
await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId)),
).toEqual([run]);
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.issueId, fixture.issueId)),
).toEqual(gates);
expect(
await db
.select({ status: agents.status })
.from(agents)
.where(eq(agents.id, fixture.agentId)),
).toEqual([{ status }]);
},
);
it.each([
"execution_owner",
"identity_revoked",
"retired_conversation",
"wrong_comment",
] as const)(
"does not treat an invokable errored agent as reviewed-chat authority after %s",
async (change) => {
const fixture = await seedWaitTurn("github");
await seedPriorCompletionReview(fixture);
await db
.update(agents)
.set({ status: "error" })
.where(eq(agents.id, fixture.agentId));
const [run] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const context = structuredClone(run!.contextSnapshot!);
if (change === "execution_owner")
await db
.update(issues)
.set({ executionRunId: null })
.where(eq(issues.id, fixture.issueId));
if (change === "identity_revoked")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.endpointId, fixture.endpointId));
if (change === "retired_conversation")
await db
.update(chatConversations)
.set({ state: "completed" })
.where(eq(chatConversations.id, fixture.conversationId));
if (change === "wrong_comment") {
context.wakeCommentId = randomUUID();
context.wakeCommentIds = [context.wakeCommentId];
}
await expect(
attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: context,
}),
).resolves.toBe(false);
},
);
it.each(["valid", "tampered_selection", "revoked_identity"])("attests only the committed Slack multiple-choice response: %s", async condition => {
const fixture = await seedAnsweredChatTurn("slack", undefined, "multi");
if (condition === "tampered_selection") {
const [action] = await db.select().from(chatActions).where(eq(chatActions.id, fixture.actionId));
await db.update(chatActions).set({ payload: { ...action!.payload, selectionValues: ["opaque-amber", "opaque-cobalt"] } }).where(eq(chatActions.id, fixture.actionId));
}
if (condition === "revoked_identity") await db.update(chatIdentityLinks).set({ status: "revoked" }).where(eq(chatIdentityLinks.principalId, fixture.principalId));
expect(await attestReviewedExternalChatRun({ db, ...fixture, contextSnapshot: fixture.context })).toBe(condition === "valid");
});
it("attests a genuine Slack mixed-question modal answer for its native continuation", async () => {
const fixture = await seedAnsweredChatTurn("slack", undefined, "form");
await expect(
attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).resolves.toBe(true);
expect(fixture.context.paperclipExternalChatQuestionResponse).toMatchObject({
schema: "paperclip.external_chat_question_response.v1",
interactionId: fixture.interactionId,
responseDeliveryId: fixture.responseDeliveryId,
sourceRunId: fixture.sourceRunId,
});
});
it("attests an accepted Slack modal answer after its opaque token expires", async () => {
const fixture = await seedAnsweredChatTurn("slack", undefined, "form");
const answeredAt = new Date(Date.now() - 2 * 60 * 60 * 1_000);
const [interaction] = await db
.update(issueThreadInteractions)
.set({ resolvedAt: answeredAt })
.where(eq(issueThreadInteractions.id, fixture.interactionId))
.returning();
const [action] = await db
.select()
.from(chatActions)
.where(eq(chatActions.id, fixture.actionId));
if (!interaction || !action) throw new Error("Expected modal answer state");
await db
.update(chatActions)
.set({
payload: {
...action.payload,
expiresAt: new Date(
answeredAt.getTime() + 60 * 60 * 1_000,
).toISOString(),
},
})
.where(eq(chatActions.id, action.id));
await db
.update(issueQuestionResponseDeliveries)
.set({
payloadSha256: questionResponseDeliveryValues(
interaction as unknown as AskUserQuestionsInteraction,
).payloadSha256,
})
.where(eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId));
await expect(
attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).resolves.toBe(true);
});
it("does not attest a Slack modal answer after current principal authorization is revoked", async () => {
const fixture = await seedAnsweredChatTurn("slack", undefined, "form");
await db
.update(companyMemberships)
.set({ status: "suspended" })
.where(
and(
eq(companyMemberships.companyId, fixture.companyId),
eq(companyMemberships.principalId, fixture.userId),
),
);
await expect(
attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).resolves.toBe(false);
});
it.each(["result", "field", "expiry"] as const)(
"does not attest a Slack modal answer with a tampered %s receipt",
async (mutation) => {
const fixture = await seedAnsweredChatTurn("slack", undefined, "form");
const [action] = await db
.select()
.from(chatActions)
.where(eq(chatActions.id, fixture.actionId));
if (!action) throw new Error("Expected modal answer action");
if (mutation === "result") {
await db
.update(chatActions)
.set({
result: {
code: "question_form_answered",
interactionId: randomUUID(),
},
})
.where(eq(chatActions.id, action.id));
} else {
const payload = structuredClone(action.payload);
if (mutation === "expiry") {
payload.expiresAt = new Date(0).toISOString();
} else {
const fields = Array.isArray(payload.fields) ? payload.fields : [];
const first = fields[0] as
| { options?: Array<{ optionId?: string }> }
| undefined;
if (!first?.options?.[0])
throw new Error("Expected modal select field");
first.options[0].optionId = "forged-option";
}
await db
.update(chatActions)
.set({ payload })
.where(eq(chatActions.id, action.id));
}
await expect(
attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).resolves.toBe(false);
expect(fixture.context).not.toHaveProperty(
"paperclipExternalChatQuestionResponse",
);
},
);
it.each([
["telegram", "register_deliverable"],
["discord", "register_deliverable"],
["telegram", "reuse_chat_attachment"],
["discord", "reuse_chat_attachment"],
["github", "register_deliverable"],
["github", "reuse_chat_attachment"],
] as const)(
"describes %s answered-question %s delivery from current authority without duplicate effects",
async (provider, tool) => {
await withAnsweredFileTool(
provider,
tool,
async ({ fixture, invoke }) => {
// Answer text is intentionally ephemeral. A prompt-shape predicate on
// the durable wake is not the authority for this file delivery mode.
expect(fixture.context.paperclipWake).not.toHaveProperty(
"questionResponse",
);
const first = await invoke();
expect(first).toMatchObject({
disposition: "applied",
fileDelivery: {
provider,
mode:
provider === "github"
? "paperclip_task_only"
: "provider_attachment",
preparationState: "prepared",
providerDeliveryConfirmed: false,
},
});
const attachments = await db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId));
expect(attachments).toHaveLength(
tool === "register_deliverable" ? 1 : 2,
);
expect(await invoke()).toEqual(first);
expect(
await db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId)),
).toEqual(attachments);
const [gate] = await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.gate.id));
expect(gate).toEqual(fixture.gate);
// Receipt replay still rechecks the real destination; a prior mode
// does not survive current reach revocation or repeat the file effect.
if (provider === "github") {
await db
.update(chatEndpointResources)
.set({ enabled: false })
.where(eq(chatEndpointResources.id, fixture.resourceId));
} else {
await db
.update(chatEndpoints)
.set({ allowDirectMessages: false })
.where(eq(chatEndpoints.id, fixture.endpointId));
}
await expect(invoke()).rejects.toThrow(
"paperclip_runner_chat_attachment_destination_denied",
);
expect(
await db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId)),
).toEqual(attachments);
},
);
},
);
it.each(["register_deliverable", "reuse_chat_attachment"] as const)(
"rejects a forged GitHub Board answer marker before %s effects",
async (tool) => {
await withAnsweredFileTool(
"github",
tool,
async ({ fixture, invoke }) => {
const attachments = await db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId));
const context = structuredClone(fixture.context);
(
context.paperclipExternalChatQuestionResponse as Record<
string,
unknown
>
).bindingSha256 = "0".repeat(64);
await db
.update(heartbeatRuns)
.set({ contextSnapshot: context })
.where(eq(heartbeatRuns.id, fixture.runId));
await expect(invoke()).rejects.toThrow(
"paperclip_runner_chat_attachment_binding_denied",
);
await expect(
db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId)),
).resolves.toEqual(attachments);
},
);
},
);
it.each([
["register_deliverable", "forged_marker"],
["reuse_chat_attachment", "forged_marker"],
["register_deliverable", "changed_generation"],
["reuse_chat_attachment", "changed_generation"],
["register_deliverable", "membership_revoked"],
["reuse_chat_attachment", "membership_revoked"],
] as const)(
"denies answered-question %s file preparation before effects for %s",
async (tool, mutation) => {
await withAnsweredFileTool(
"telegram",
tool,
async ({ fixture, invoke }) => {
const attachments = await db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId));
const [before] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
if (mutation === "forged_marker") {
const context = structuredClone(fixture.context);
const marker =
context.paperclipExternalChatQuestionResponse as Record<
string,
unknown
>;
marker.bindingSha256 = "0".repeat(64);
await db
.update(heartbeatRuns)
.set({ contextSnapshot: context })
.where(eq(heartbeatRuns.id, fixture.runId));
} else if (mutation === "changed_generation") {
await db
.update(chatConversations)
.set({ sessionGeneration: 2 })
.where(eq(chatConversations.id, fixture.conversationId));
} else {
await db
.update(companyMemberships)
.set({ status: "suspended" })
.where(
and(
eq(companyMemberships.companyId, fixture.companyId),
eq(companyMemberships.principalId, fixture.userId),
),
);
}
await expect(invoke()).rejects.toThrow(
mutation === "membership_revoked"
? "paperclip_runner_chat_attachment_principal_denied"
: "paperclip_runner_chat_attachment_binding_denied",
);
expect(
await db
.select()
.from(issueAttachments)
.where(eq(issueAttachments.issueId, fixture.issueId)),
).toEqual(attachments);
const [after] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(after?.resultJson).toEqual(before?.resultJson);
},
);
},
);
async function seedSequentialQuestionChain(depth = 2) {
const fixture = await seedAnsweredChatTurn();
const parents: Array<Awaited<ReturnType<typeof seedAnsweredChatTurn>>> = [];
let cursor = fixture;
for (let index = 1; index < depth; index += 1) {
cursor = await seedAnsweredChatTurn("telegram", {
fixture: { ...fixture, runId: cursor.sourceRunId },
gate: fixture.gate,
});
parents.unshift(cursor);
}
for (const parent of parents) {
await db
.update(issues)
.set({ executionRunId: parent.runId })
.where(eq(issues.id, fixture.issueId));
await attestAnswer(parent);
await db
.update(heartbeatRuns)
.set({ status: "succeeded" })
.where(eq(heartbeatRuns.id, parent.runId));
}
await db
.update(issues)
.set({ executionRunId: fixture.runId })
.where(eq(issues.id, fixture.issueId));
return { fixture, parents };
}
it.each([1, 3])(
"reports only the latest durable answer time for a %i-question chain after a long wait",
async (depth) => {
const { fixture, parents } = await seedSequentialQuestionChain(depth);
const answeredAt = new Date(Date.now() + 7_200_000);
const originalCreatedAt = new Date(answeredAt.getTime() - 14_400_000);
for (const parent of parents) {
const [earlier] = await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, parent.interactionId));
expect(answeredAt.getTime()).toBeGreaterThan(
earlier!.resolvedAt!.getTime(),
);
}
await db
.update(issueComments)
.set({ createdAt: originalCreatedAt })
.where(eq(issueComments.id, fixture.commentId));
const [interaction] = await db
.update(issueThreadInteractions)
.set({ resolvedAt: answeredAt })
.where(eq(issueThreadInteractions.id, fixture.interactionId))
.returning();
const delivery = questionResponseDeliveryValues(
interaction as unknown as AskUserQuestionsInteraction,
);
await db
.update(issueQuestionResponseDeliveries)
.set({ payloadSha256: delivery.payloadSha256 })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
// Neither a caller timestamp nor an untrusted marker field is the source.
fixture.context.answeredAtMs = 1;
fixture.context.paperclipExternalChatQuestionResponse = {
answeredAtMs: 2,
};
const onQuestionResponseAttested = vi.fn();
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
onQuestionResponseAttested,
}),
).toBe(true);
expect(onQuestionResponseAttested).toHaveBeenCalledExactlyOnceWith(
answeredAt.getTime(),
);
expect(fixture.context.sourceCommentId).toBe(fixture.commentId);
expect(fixture.context.wakeCommentIds).toEqual([fixture.commentId]);
expect(
fixture.context.paperclipExternalChatQuestionResponse,
).not.toHaveProperty("answeredAtMs");
const [comment] = await db
.select()
.from(issueComments)
.where(eq(issueComments.id, fixture.commentId));
expect(comment!.createdAt).toEqual(originalCreatedAt);
},
);
it.each(["forged_source", "stale_answer", "unattested_principal"] as const)(
"does not expose an answered-question timestamp for %s",
async (kind) => {
const fixture = await seedAnsweredChatTurn();
if (kind === "forged_source") {
fixture.context.sourceRunId = randomUUID();
await db
.update(heartbeatRuns)
.set({ contextSnapshot: fixture.context })
.where(eq(heartbeatRuns.id, fixture.runId));
}
if (kind === "stale_answer")
await db
.update(issueQuestionResponseDeliveries)
.set({ payloadSha256: "0".repeat(64) })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
if (kind === "unattested_principal")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
fixture.context.answeredAtMs = 1;
const onQuestionResponseAttested = vi.fn();
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
onQuestionResponseAttested,
}),
).toBe(false);
expect(onQuestionResponseAttested).not.toHaveBeenCalled();
},
);
it("does not expose an answered-question timestamp before its attestation transaction commits", async () => {
const fixture = await seedAnsweredChatTurn();
const onQuestionResponseAttested = vi.fn();
let validatedInsideTransaction = false;
const rolledBackDb = {
transaction: async (operation: (tx: typeof db) => Promise<unknown>) =>
db.transaction(async (tx) => {
await operation(tx as unknown as typeof db);
validatedInsideTransaction = true;
throw new Error("timing_attestation_test_rollback");
}),
} as unknown as typeof db;
await expect(
attestReviewedExternalChatRun({
db: rolledBackDb,
...fixture,
contextSnapshot: fixture.context,
onQuestionResponseAttested,
}),
).rejects.toThrow("timing_attestation_test_rollback");
expect(validatedInsideTransaction).toBe(true);
expect(onQuestionResponseAttested).not.toHaveBeenCalled();
});
it("authorizes sequential chat questions through exact durable parents and preserves the original request", async () => {
const { fixture, parents } = await seedSequentialQuestionChain(3);
await attestAnswer(fixture);
const resolved = await resolveExternalChatQuestionResponse(
db,
fixture,
fixture.context,
"read",
);
expect(resolved?.interactionIds).toEqual([
...parents.map((parent) => parent.interactionId),
fixture.interactionId,
]);
expect(resolved?.marker.sourceCommentId).toBe(fixture.commentId);
expect(fixture.context.source).toBe("issue.interaction.respond");
expect(fixture.context.wakeCommentIds).toEqual([fixture.commentId]);
const responses = await materializeExternalChatQuestionResponseInput({
db, binding: fixture, contextSnapshot: fixture.context,
});
expect(responses.map((response) => response.interactionId)).toEqual(resolved!.interactionIds);
const resultJson = await finishReviewResponse(fixture);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.gate.id)),
).toEqual([expect.objectContaining({ status: "pending" })]);
});
it.each(["execution_owner", "agent_paused", "membership_revoked"] as const)(
"rechecks current sequential answer input authority after attestation: %s",
async (kind) => {
const { fixture } = await seedSequentialQuestionChain();
await attestAnswer(fixture);
if (kind === "execution_owner")
await db
.update(issues)
.set({ executionRunId: fixture.sourceRunId })
.where(eq(issues.id, fixture.issueId));
if (kind === "agent_paused")
await db
.update(agents)
.set({ status: "paused" })
.where(eq(agents.id, fixture.agentId));
if (kind === "membership_revoked")
await db
.update(companyMemberships)
.set({ status: "suspended" })
.where(
and(
eq(companyMemberships.companyId, fixture.companyId),
eq(companyMemberships.principalId, fixture.userId),
),
);
await expect(
materializeExternalChatQuestionResponseInput({
db,
binding: fixture,
contextSnapshot: fixture.context,
}),
).rejects.toThrow(
kind === "membership_revoked"
? "paperclip_runner_chat_attachment_principal_denied"
: "reviewed_chat_execution_binding_not_authorized",
);
},
);
it("materializes a sequential answer chain atomically with authorization before a coherent ancestor rewrite", async () => {
const { fixture, parents } = await seedSequentialQuestionChain();
await attestAnswer(fixture);
const parent = parents[0]!;
let reached!: () => void;
let release!: () => void;
const ready = new Promise<void>((resolve) => {
reached = resolve;
});
const released = new Promise<void>((resolve) => {
release = resolve;
});
const original =
nativeInteractionBridge.materializeNativeInteractionResponses;
const spy = vi
.spyOn(nativeInteractionBridge, "materializeNativeInteractionResponses")
.mockImplementationOnce(async (input) => {
reached();
await released;
return original(input);
});
const prompt = materializeExternalChatQuestionResponseInput({
db,
binding: fixture,
contextSnapshot: fixture.context,
});
let mutation: Promise<void> | null = null;
let mutationPid = 0;
try {
await Promise.race([
ready,
prompt.then(() => {
throw new Error("materialization_barrier_not_reached");
}),
]);
mutation = db.transaction(async (tx) => {
const [backend] = await tx.execute(sql`select pg_backend_pid() as pid`);
mutationPid = Number(backend!.pid);
const [interaction] = await tx
.update(issueThreadInteractions)
.set({
result: {
version: 1,
answers: [{ questionId: "color", optionIds: ["amber"] }],
},
})
.where(eq(issueThreadInteractions.id, parent.interactionId))
.returning();
const [action] = await tx
.select()
.from(chatActions)
.where(eq(chatActions.id, parent.actionId));
await tx
.update(chatActions)
.set({ payload: { ...action!.payload, optionId: "amber" } })
.where(eq(chatActions.id, parent.actionId));
await tx
.update(issueQuestionResponseDeliveries)
.set({
payloadSha256: questionResponseDeliveryValues(
interaction! as unknown as AskUserQuestionsInteraction,
).payloadSha256,
})
.where(
eq(issueQuestionResponseDeliveries.id, parent.responseDeliveryId),
);
});
await vi.waitFor(async () => {
expect(mutationPid).toBeGreaterThan(0);
const [waiting] = await db.execute(
sql`select exists(select 1 from pg_locks where pid = ${mutationPid} and not granted) as waiting`,
);
expect(waiting!.waiting).toBe(true);
});
release();
const captured = await prompt;
await mutation;
expect(captured.map((response) => response.interactionId)).toEqual([
parent.interactionId,
fixture.interactionId,
]);
expect(JSON.stringify(captured)).toContain("Cobalt");
expect(JSON.stringify(captured)).not.toContain("Amber");
await expect(
materializeExternalChatQuestionResponseInput({
db,
binding: fixture,
contextSnapshot: fixture.context,
}),
).rejects.toThrow("reviewed_chat_execution_binding_not_authorized");
} finally {
release();
await Promise.allSettled([prompt, ...(mutation ? [mutation] : [])]);
spy.mockRestore();
}
});
it.each([
"missing_parent_marker",
"missing_parent_delivery",
"tampered_parent_answer",
"rewritten_parent_answer_and_receipt",
"duplicate_parent_action",
"revoked_parent_principal",
"wrong_parent_actor",
"source_cycle",
"different_parent_issue",
] as const)(
"rejects an unauthenticated sequential chat question chain: %s",
async (kind) => {
const { fixture, parents } = await seedSequentialQuestionChain();
const parent = parents[0]!;
if (kind === "missing_parent_marker") {
const context = { ...parent.context };
delete context.paperclipExternalChatQuestionResponse;
await db
.update(heartbeatRuns)
.set({ contextSnapshot: context })
.where(eq(heartbeatRuns.id, parent.runId));
}
if (kind === "missing_parent_delivery")
await db
.delete(issueQuestionResponseDeliveries)
.where(
eq(issueQuestionResponseDeliveries.id, parent.responseDeliveryId),
);
if (kind === "tampered_parent_answer")
await db
.update(issueThreadInteractions)
.set({
result: {
version: 1,
answers: [{ questionId: "color", optionIds: ["amber"] }],
},
})
.where(eq(issueThreadInteractions.id, parent.interactionId));
if (kind === "rewritten_parent_answer_and_receipt") {
const [interaction] = await db
.update(issueThreadInteractions)
.set({
result: {
version: 1,
answers: [{ questionId: "color", optionIds: ["amber"] }],
},
})
.where(eq(issueThreadInteractions.id, parent.interactionId))
.returning();
const [action] = await db
.select()
.from(chatActions)
.where(eq(chatActions.id, parent.actionId));
await db
.update(chatActions)
.set({ payload: { ...action!.payload, optionId: "amber" } })
.where(eq(chatActions.id, parent.actionId));
await db
.update(issueQuestionResponseDeliveries)
.set({
payloadSha256: questionResponseDeliveryValues(
interaction! as unknown as AskUserQuestionsInteraction,
).payloadSha256,
})
.where(
eq(issueQuestionResponseDeliveries.id, parent.responseDeliveryId),
);
}
if (kind === "duplicate_parent_action") {
const [action] = await db
.select()
.from(chatActions)
.where(eq(chatActions.id, parent.actionId));
await db
.insert(chatActions)
.values({
...action!,
id: randomUUID(),
providerActionId: "duplicate-parent-response",
});
}
if (kind === "revoked_parent_principal")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
if (kind === "wrong_parent_actor")
await db
.update(agentWakeupRequests)
.set({ requestedByActorId: "another-user" })
.where(eq(agentWakeupRequests.id, parent.wakeId));
if (kind === "different_parent_issue") {
// Deliberately corrupt historical data; normal writes reject this rebind.
await db.transaction(async (tx) => {
await tx.execute(sql`set local session_replication_role = replica`);
await tx.update(heartbeatRuns).set({ nativeIssueId: fixture.sourceRunId }).where(eq(heartbeatRuns.id, parent.runId));
});
}
if (kind === "source_cycle") {
const [wake] = await db
.select()
.from(agentWakeupRequests)
.where(eq(agentWakeupRequests.id, parent.wakeId));
await db
.update(agentWakeupRequests)
.set({ payload: { ...wake!.payload, sourceRunId: fixture.runId } })
.where(eq(agentWakeupRequests.id, parent.wakeId));
await db
.update(issueThreadInteractions)
.set({ sourceRunId: fixture.runId })
.where(eq(issueThreadInteractions.id, parent.interactionId));
await db
.update(issueQuestionResponseDeliveries)
.set({ sourceRunId: fixture.runId })
.where(
eq(issueQuestionResponseDeliveries.id, parent.responseDeliveryId),
);
await db
.update(heartbeatRuns)
.set({
contextSnapshot: { ...parent.context, sourceRunId: fixture.runId },
})
.where(eq(heartbeatRuns.id, parent.runId));
}
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).toBe(false);
expect(
fixture.context.paperclipExternalChatQuestionResponse,
).toBeUndefined();
},
);
it("bounds sequential question ancestry without silently dropping earlier answers", async () => {
const { fixture, parents } = await seedSequentialQuestionChain(9);
const deepestAllowed = parents.at(-1)!;
expect(
(
await resolveExternalChatQuestionResponse(
db,
deepestAllowed,
deepestAllowed.context,
"read",
)
)?.interactionIds,
).toHaveLength(8);
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).toBe(false);
});
it("revalidates sequential question ancestors before publishing the later answer", async () => {
const { fixture, parents } = await seedSequentialQuestionChain();
await attestAnswer(fixture);
const resultJson = await finishReviewResponse(fixture);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
await db
.update(agentWakeupRequests)
.set({ requestedByActorId: "another-user" })
.where(eq(agentWakeupRequests.id, parents[0]!.wakeId));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(false);
});
it.each(["telegram", "discord"] as const)(
"retains authenticated %s answer continuation presentation without resolving prior review",
async (provider) => {
const fixture = await seedAnsweredChatTurn(provider);
await attestAnswer(fixture);
const resultJson = await finishReviewResponse(fixture);
expect(resultJson.externalChatReviewPresentation).toMatchObject({
gateId: fixture.gate.id,
});
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
const [gate] = await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.gate.id));
expect(gate!.status).toBe("pending");
},
);
it.each([
"wake_actor",
"different_responder",
"revoked_link",
"wrong_option",
"wrong_digest",
"wrong_target",
"wrong_wake_target",
"missing_wake_receipt",
"unprocessed_action",
"duplicate_action",
"source_issue",
"source_batch",
"execution_owner",
] as const)(
"does not attest an unbound answered-chat continuation: %s",
async (kind) => {
const fixture = await seedAnsweredChatTurn();
if (kind === "wake_actor")
await db
.update(agentWakeupRequests)
.set({ requestedByActorId: "another-user" })
.where(eq(agentWakeupRequests.id, fixture.wakeId));
if (kind === "different_responder") {
// A legitimate linked second responder must not inherit the original author's scope.
const principalId = randomUUID();
await db
.insert(chatExternalPrincipals)
.values({
id: principalId,
companyId: fixture.companyId,
provider: "telegram",
providerAccountId: "telegram-bot",
externalId: "second-user",
kind: "user",
});
await db
.insert(chatIdentityLinks)
.values({
companyId: fixture.companyId,
endpointId: fixture.endpointId,
principalId,
paperclipUserId: "second-user",
status: "linked",
});
await db
.insert(companyMemberships)
.values({
companyId: fixture.companyId,
principalType: "user",
principalId: "second-user",
status: "active",
membershipRole: "member",
});
await db
.update(chatActions)
.set({ principalId })
.where(eq(chatActions.id, fixture.actionId));
await db
.update(issueThreadInteractions)
.set({ resolvedByUserId: "second-user" })
.where(eq(issueThreadInteractions.id, fixture.interactionId));
await db
.update(agentWakeupRequests)
.set({ requestedByActorId: "second-user" })
.where(eq(agentWakeupRequests.id, fixture.wakeId));
}
if (kind === "revoked_link")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
if (kind === "wrong_option") {
const [action] = await db
.select()
.from(chatActions)
.where(eq(chatActions.id, fixture.actionId));
await db
.update(chatActions)
.set({ payload: { ...action!.payload, optionId: "amber" } })
.where(eq(chatActions.id, fixture.actionId));
}
if (kind === "wrong_digest")
await db
.update(issueQuestionResponseDeliveries)
.set({ payloadSha256: "b".repeat(64) })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
if (kind === "wrong_target")
await db
.update(issueQuestionResponseDeliveries)
.set({ targetRunId: fixture.sourceRunId })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
if (kind === "wrong_wake_target")
await db
.update(agentWakeupRequests)
.set({ runId: fixture.sourceRunId })
.where(eq(agentWakeupRequests.id, fixture.wakeId));
if (kind === "missing_wake_receipt")
await db
.update(heartbeatRuns)
.set({ wakeupRequestId: null })
.where(eq(heartbeatRuns.id, fixture.runId));
if (kind === "unprocessed_action")
await db
.update(chatActions)
.set({ status: "issued" })
.where(eq(chatActions.id, fixture.actionId));
if (kind === "duplicate_action") {
const [action] = await db
.select()
.from(chatActions)
.where(eq(chatActions.id, fixture.actionId));
await db
.insert(chatActions)
.values({
...action!,
id: randomUUID(),
providerActionId: "duplicate-response",
});
}
if (kind === "source_issue" || kind === "source_batch") {
const [source] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.sourceRunId));
await db
.update(heartbeatRuns)
.set({
contextSnapshot: {
...source!.contextSnapshot,
...(kind === "source_issue"
? { issueId: randomUUID() }
: { wakeCommentIds: [randomUUID(), fixture.commentId] }),
},
})
.where(eq(heartbeatRuns.id, fixture.sourceRunId));
}
if (kind === "execution_owner")
await db
.update(issues)
.set({ executionRunId: fixture.sourceRunId })
.where(eq(issues.id, fixture.issueId));
expect(
await attestReviewedExternalChatRun({
db,
...fixture,
contextSnapshot: fixture.context,
}),
).toBe(false);
expect(
fixture.context.paperclipExternalChatQuestionResponse,
).toBeUndefined();
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, fixture.gate.id)),
).toEqual([expect.objectContaining({ status: "pending" })]);
},
);
it.each([
"revoked_link",
"revoked_membership",
"demoted_membership",
"changed_generation",
"changed_answer",
"changed_gate",
] as const)(
"revalidates answered-chat presentation at dispatch: %s",
async (kind) => {
const fixture = await seedAnsweredChatTurn();
await attestAnswer(fixture);
const resultJson = await finishReviewResponse(fixture);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
if (kind === "revoked_link")
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
if (kind === "revoked_membership")
await db
.update(companyMemberships)
.set({ status: "suspended" })
.where(
and(
eq(companyMemberships.companyId, fixture.companyId),
eq(companyMemberships.principalId, fixture.userId),
),
);
if (kind === "demoted_membership")
await db
.update(companyMemberships)
.set({ membershipRole: "viewer" })
.where(
and(
eq(companyMemberships.companyId, fixture.companyId),
eq(companyMemberships.principalId, fixture.userId),
),
);
if (kind === "changed_generation")
await db
.update(chatConversations)
.set({ sessionGeneration: 2 })
.where(eq(chatConversations.id, fixture.conversationId));
if (kind === "changed_answer")
await db
.update(issueThreadInteractions)
.set({
result: {
version: 1,
answers: [{ questionId: "color", optionIds: ["amber"] }],
},
})
.where(eq(issueThreadInteractions.id, fixture.interactionId));
if (kind === "changed_gate")
await db
.update(issueThreadInteractions)
.set({ status: "accepted" })
.where(eq(issueThreadInteractions.id, fixture.gate.id));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(false);
},
);
it("does not register a fallback after an answered-chat principal is revoked without a review gate", async () => {
const fixture = await seedAnsweredChatTurn();
await attestAnswer(fixture);
await db
.update(issueThreadInteractions)
.set({ status: "accepted" })
.where(eq(issueThreadInteractions.id, fixture.gate.id));
await db
.update(issues)
.set({ status: "in_progress" })
.where(eq(issues.id, fixture.issueId));
await db
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
await finishReviewResponse(fixture);
expect(
await db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.runId, fixture.runId)),
).toEqual([
expect.objectContaining({
reasonCode: "external_chat_response_wait_authorization_lost",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
expect(
await db
.select()
.from(agentWakeupRequests)
.where(eq(agentWakeupRequests.companyId, fixture.companyId)),
).toHaveLength(1);
});
it("accepts the exact admitted answer while its post-wakeup delivery receipt is still finishing", async () => {
const fixture = await seedAnsweredChatTurn();
await db
.update(issueQuestionResponseDeliveries)
.set({ status: "delivering", deliveryMode: null, targetRunId: null })
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
await attestAnswer(fixture);
await db
.update(issueQuestionResponseDeliveries)
.set({
status: "fallback_queued",
deliveryMode: "wake_fallback",
targetRunId: fixture.runId,
})
.where(
eq(issueQuestionResponseDeliveries.id, fixture.responseDeliveryId),
);
const resultJson = await finishReviewResponse(fixture);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
});
it("takes the answered-chat identity advisory before identity rows during concurrent revocation", async () => {
const fixture = await seedAnsweredChatTurn();
await attestAnswer(fixture);
let release!: () => void;
let acquired!: () => void;
let revoke!: () => void;
let confirmRevoked!: () => void;
let rejectRevoked!: (error: unknown) => void;
const released = new Promise<void>((resolve) => {
release = resolve;
});
const ready = new Promise<void>((resolve) => {
acquired = resolve;
});
const shouldRevoke = new Promise<void>((resolve) => {
revoke = resolve;
});
const revoked = new Promise<void>((resolve, reject) => {
confirmRevoked = resolve;
rejectRevoked = reject;
});
const holder = db.transaction(async (tx) => {
await tx.execute(
sql`select pg_advisory_xact_lock(hashtextextended(${`chat-identity:${fixture.companyId}:${fixture.principalId}`}, 0))`,
);
acquired();
await shouldRevoke;
try {
await tx
.select()
.from(chatIdentityLinks)
.where(eq(chatIdentityLinks.principalId, fixture.principalId))
.for("update", { noWait: true });
await tx
.update(chatIdentityLinks)
.set({ status: "revoked" })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
confirmRevoked();
} catch (error) {
rejectRevoked(error);
throw error;
} finally {
await released;
}
});
await ready;
let readerPid = 0;
const reader = db
.transaction(async (tx) => {
const [backend] = await tx.execute(sql`select pg_backend_pid() as pid`);
readerPid = Number(backend!.pid);
return authorizeChatConversationForBoundRun(
tx as unknown as typeof db,
fixture,
fixture.context,
);
})
.then(
() => "unexpectedly_authorized",
(error: Error) => error.message,
);
try {
await vi.waitFor(async () => {
expect(readerPid).toBeGreaterThan(0);
const [waiting] = await db.execute(
sql`select exists(select 1 from pg_locks where pid = ${readerPid} and locktype = 'advisory' and not granted) as waiting`,
);
expect(waiting!.waiting).toBe(true);
});
revoke();
await Promise.race([revoked, holder]);
} finally {
revoke();
release();
await Promise.allSettled([holder, reader]);
}
expect(await reader).toBe(
"paperclip_runner_chat_attachment_binding_denied",
);
});
async function seedPriorCompletionReview(
fixture: Awaited<ReturnType<typeof seedWaitTurn>>,
) {
const [current] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const [accepted] = await db
.select()
.from(nativeRunResults)
.where(eq(nativeRunResults.runId, fixture.runId));
const runId = randomUUID();
const sessionId = randomUUID();
const runnerInstanceId = randomUUID();
await db.insert(heartbeatRuns).values({
id: runId,
companyId: fixture.companyId,
agentId: fixture.agentId,
status: "running",
runtimeMode: "native",
nativeIssueId: fixture.issueId,
nativeSessionId: sessionId,
runnerInstanceId,
completionContractId: current!.completionContractId,
completionContractSha256: current!.completionContractSha256,
contextSnapshot: {},
createdAt: new Date(Date.now() - 60_000),
});
await db
.update(issues)
.set({ executionRunId: runId })
.where(eq(issues.id, fixture.issueId));
const port = new PaperclipControlPlanePort(db, {
companyId: fixture.companyId,
issueId: fixture.issueId,
agentId: fixture.agentId,
runId,
sessionId,
completionContractId: current!.completionContractId!,
completionContractSha256: current!.completionContractSha256!,
sourceInstanceId: runnerInstanceId,
controlPlaneSourceInstanceId: `review-control-${runId}`,
});
await port.openRun({
identity: {
companyId: fixture.companyId,
issueId: fixture.issueId,
agentId: fixture.agentId,
runId,
sessionId,
},
backendKind: "mock",
sourceInstanceId: runnerInstanceId,
});
const result = {
...(accepted!.resultJson.result as PrpStructuredRunResult),
reportedWorkDisposition: "needs_review" as const,
};
delete result.continuation;
result.attentionRequests = [{ kind: "review", ownerClass: "human", summary: "Approve the prepared response before continuing." }];
const terminal = {
...(accepted!.resultJson.terminal as PrpTerminalState),
reportedWorkDisposition: "needs_review" as const,
};
await port.completeRun({
result,
terminal,
callerResultId: `prior-review-${runId}`,
});
await finalizeNativeRun({
db,
runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
const [gate] = await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.sourceRunId, runId));
expect(gate).toMatchObject({
kind: "request_confirmation",
status: "pending",
createdByAgentId: null,
createdByUserId: null,
});
await db
.update(issues)
.set({ executionRunId: fixture.runId })
.where(eq(issues.id, fixture.issueId));
await db
.update(heartbeatRuns)
.set({ startedAt: new Date(gate!.createdAt.getTime() + 1) })
.where(eq(heartbeatRuns.id, fixture.runId));
return gate!;
}
async function finishReviewResponse(
fixture: Awaited<ReturnType<typeof seedWaitTurn>>,
) {
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
const [run] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
return run!.resultJson!;
}
it.each(["telegram", "discord"] as const)(
"presents the exact %s response without resolving a pre-existing native completion review",
async (provider) => {
const fixture = await seedWaitTurn(provider);
const gate = await seedPriorCompletionReview(fixture);
const selectedComment = await issueService(db).addComment(
fixture.issueId,
"Prepared exactly these files for this chat response.",
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "paperclip_runner_protocol" },
);
const selected = [];
for (const [originalFilename, contentType] of [
["original-cat.png", "image/png"],
["original-notes.txt", "text/plain"],
]) {
selected.push(
await issueService(db).createAttachment({
issueId: fixture.issueId,
issueCommentId: selectedComment.id,
provider: "local_disk",
objectKey: `issues/${fixture.issueId}/${originalFilename}`,
contentType: contentType!,
byteSize: 128,
sha256: "a".repeat(64),
originalFilename,
createdByAgentId: fixture.agentId,
createdByRunId: fixture.runId,
}),
);
}
expect(
await db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId)),
).toEqual([]);
const resultJson = await finishReviewResponse(fixture);
expect(resultJson).toMatchObject({
finalizationPhase: "committed",
finalizationReasonCode: "governed_response_waiting",
externalChatReviewPresentation: {
runId: fixture.runId,
gateId: gate.id,
gateDecisionId: (gate.payload as { target: { revisionId: string } })
.target.revisionId,
},
});
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
expect(
await resolveChatRunPresentationAuthorizationReason(db, fixture),
).toBe("allow_chat_run_presentation");
const response = resolveHeartbeatRunResponse({
resultJson,
preferFinalResponseOverExistingComment: true,
externalChatResponseWakeSummaryAuthorized: true,
externalChatReviewResponseSummaryAuthorized: true,
});
expect(response.text).toBe(
"The requested photo is prepared. I will wait for your next message.",
);
const first = await issueService(db).addComment(
fixture.issueId,
response.text!,
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "allow_chat_run_presentation" },
);
const second = await issueService(db).addComment(
fixture.issueId,
response.text!,
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "allow_chat_run_presentation" },
);
expect(second.id).toBe(first.id);
const publications = await db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId));
expect(publications).toHaveLength(3);
expect(publications).toEqual(
expect.arrayContaining(
selected.map((attachment) =>
expect.objectContaining({
commentId: selectedComment.id,
state: "pending",
idempotencyKey: `attachment:${attachment.id}:${fixture.endpointId}`,
payload: expect.objectContaining({
attachmentIds: [attachment.id],
}),
}),
),
),
);
await expect(
issueService(db).addComment(
fixture.issueId,
"Not the accepted summary",
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "allow_chat_run_presentation" },
),
).rejects.toMatchObject({
status: 409,
details: { code: "chat_review_response_presentation_denied" },
});
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, gate.id)),
).toEqual([
expect.objectContaining({ status: "pending", resolvedAt: null }),
]);
expect(
await db.select().from(issues).where(eq(issues.id, fixture.issueId)),
).toEqual([expect.objectContaining({ status: "in_review" })]);
expect(
await db
.select()
.from(agentWakeupRequests)
.where(eq(agentWakeupRequests.companyId, fixture.companyId)),
).toEqual([]);
},
);
it.each([
"approval",
"agent_gate",
"new_attention",
"semantic_attention",
"revoked_endpoint",
] as const)(
"does not mint a review-presentation grant for %s",
async (kind) => {
const fixture = await seedWaitTurn(
"telegram",
kind === "semantic_attention"
? [
{
kind: "approval",
summary: "Approve a separate action",
ownerClass: "human",
},
]
: [],
);
const gate = await seedPriorCompletionReview(fixture);
if (kind === "approval") {
const [approval] = await db
.insert(approvals)
.values({
companyId: fixture.companyId,
type: "hire_agent",
status: "pending",
payload: {},
})
.returning();
await db.insert(issueApprovals).values({
companyId: fixture.companyId,
issueId: fixture.issueId,
approvalId: approval!.id,
});
} else if (kind === "agent_gate") {
await db
.update(issueThreadInteractions)
.set({ createdByAgentId: fixture.agentId })
.where(eq(issueThreadInteractions.id, gate.id));
} else if (kind === "new_attention") {
await db.insert(issueThreadInteractions).values({
companyId: fixture.companyId,
issueId: fixture.issueId,
kind: "request_confirmation",
status: "pending",
sourceRunId: fixture.runId,
createdByAgentId: fixture.agentId,
title: "Current confirmation",
payload: gate.payload,
});
} else if (kind === "revoked_endpoint") {
await db
.update(chatEndpoints)
.set({ status: "paused" })
.where(eq(chatEndpoints.id, fixture.endpointId));
}
const resultJson = await finishReviewResponse(fixture);
expect(resultJson.externalChatReviewPresentation).toBeNull();
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(false);
expect(
await resolveChatRunPresentationAuthorizationReason(db, fixture),
).toBe("internal_agent_write");
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, gate.id)),
).toEqual([expect.objectContaining({ status: "pending" })]);
},
);
it.each(["next_run", "concurrent_other_issue"] as const)(
"keeps a committed successful response authorized after runtime failure in %s",
async (failureScope) => {
const fixture = await seedWaitTurn("discord");
const gate = await seedPriorCompletionReview(fixture);
const resultJson = await finishReviewResponse(fixture);
const laterIssueId =
failureScope === "next_run" ? fixture.issueId : randomUUID();
const laterRunId = randomUUID();
if (laterIssueId !== fixture.issueId) {
await db
.insert(issues)
.values({
id: laterIssueId,
companyId: fixture.companyId,
title: "Concurrent task",
status: "in_progress",
assigneeAgentId: fixture.agentId,
});
}
await db
.insert(heartbeatRuns)
.values({
id: laterRunId,
companyId: fixture.companyId,
agentId: fixture.agentId,
runtimeMode: "native",
nativeIssueId: laterIssueId,
status: "running",
startedAt: new Date(),
});
await db
.update(issues)
.set({ executionRunId: laterRunId })
.where(eq(issues.id, laterIssueId));
const failRun = async (tx: typeof db) => {
await tx
.update(heartbeatRuns)
.set({
status: "failed",
errorCode: "adapter_failed",
finishedAt: new Date(),
})
.where(eq(heartbeatRuns.id, laterRunId));
await tx
.update(issues)
.set({ executionRunId: null })
.where(eq(issues.id, laterIssueId));
// This is finalizeAgentStatus's failed-last-running-run projection,
// not a user pause, termination, or permission change.
await tx
.update(agents)
.set({ status: "error", errorReason: "Later runner startup failed" })
.where(eq(agents.id, fixture.agentId));
};
if (failureScope === "concurrent_other_issue") {
let ready!: () => void;
let release!: () => void;
const observed = new Promise<void>((resolve) => {
ready = resolve;
});
const released = new Promise<void>((resolve) => {
release = resolve;
});
const holder = db.transaction(async (tx) => {
await failRun(tx as unknown as typeof db);
ready();
await released;
});
await observed;
try {
const attempt = await db
.transaction((tx) =>
authorizeNativeChatReviewPresentation(
tx as unknown as typeof db,
{ ...fixture, resultJson },
"nonblocking",
),
)
.then(
(value) => ({ value }),
(error: unknown) => ({ error }),
);
expect(attempt).toHaveProperty("error");
if ("error" in attempt)
expect(
isExternalChatWaitAuthorizationContention(attempt.error),
).toBe(true);
} finally {
release();
await holder;
}
} else {
await db.transaction((tx) => failRun(tx as unknown as typeof db));
}
expect(
await db.transaction((tx) =>
authorizeNativeChatReviewPresentation(
tx as unknown as typeof db,
{ ...fixture, resultJson },
"nonblocking",
),
),
).toBe(true);
const summary = (resultJson.nativeResult as { summary: string }).summary;
const append = () =>
issueService(db).addComment(
fixture.issueId,
summary,
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "allow_chat_run_presentation" },
);
const first = await append();
expect((await append()).id).toBe(first.id);
expect(
await db
.select()
.from(chatPublications)
.where(eq(chatPublications.commentId, first.id)),
).toHaveLength(1);
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.id, gate.id)),
).toEqual([gate]);
expect(
await db
.select({ status: agents.status })
.from(agents)
.where(eq(agents.id, fixture.agentId)),
).toEqual([{ status: "error" }]);
expect(
await db
.select({ status: heartbeatRuns.status })
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, laterRunId)),
).toEqual([{ status: "failed" }]);
expect(
await db
.select()
.from(agentWakeupRequests)
.where(eq(agentWakeupRequests.companyId, fixture.companyId)),
).toEqual([]);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
runId: laterRunId,
issueId: laterIssueId,
resultJson,
}),
).toBe(false);
await db
.update(companyMemberships)
.set({ status: "suspended" })
.where(eq(companyMemberships.principalId, fixture.userId));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(false);
},
);
it.each([
"gate_changed",
"gate_policy_changed",
"failed_run",
"status_changed",
"revoked_endpoint",
"revoked_principal",
"agent_paused",
"agent_budget_paused",
"agent_terminated",
"agent_pending_approval",
"spoofed_marker",
"changed_summary",
"changed_context",
"wrong_destination",
] as const)(
"rejects a stale or forged committed response grant: %s",
async (kind) => {
const fixture = await seedWaitTurn();
const gate = await seedPriorCompletionReview(fixture);
const resultJson = await finishReviewResponse(fixture);
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
if (kind === "gate_changed")
await db
.update(issueThreadInteractions)
.set({ status: "accepted" })
.where(eq(issueThreadInteractions.id, gate.id));
if (kind === "status_changed")
await db
.update(issues)
.set({ status: "cancelled" })
.where(eq(issues.id, fixture.issueId));
if (kind === "gate_policy_changed")
await db
.update(issueThreadInteractions)
.set({ effectiveResolverPolicy: "anyone" })
.where(eq(issueThreadInteractions.id, gate.id));
if (kind === "failed_run")
await db
.update(heartbeatRuns)
.set({ status: "failed" })
.where(eq(heartbeatRuns.id, fixture.runId));
if (kind === "revoked_endpoint")
await db
.update(chatEndpoints)
.set({ status: "paused" })
.where(eq(chatEndpoints.id, fixture.endpointId));
if (kind === "revoked_principal")
await db
.update(companyMemberships)
.set({ status: "suspended" })
.where(eq(companyMemberships.principalId, fixture.userId));
if (
[
"agent_paused",
"agent_budget_paused",
"agent_terminated",
"agent_pending_approval",
].includes(kind)
)
await db
.update(agents)
.set({
status:
kind === "agent_terminated"
? "terminated"
: kind === "agent_pending_approval"
? "pending_approval"
: "paused",
pauseReason:
kind === "agent_budget_paused"
? "budget"
: kind === "agent_paused"
? "manual"
: null,
})
.where(eq(agents.id, fixture.agentId));
if (kind === "spoofed_marker")
resultJson.externalChatReviewPresentation = {
...(resultJson.externalChatReviewPresentation as object),
gateId: randomUUID(),
};
if (kind === "changed_summary")
resultJson.nativeResult = {
...(resultJson.nativeResult as object),
summary: "Unapproved replacement prose",
};
if (kind === "changed_context")
await db
.update(heartbeatRuns)
.set({ contextSnapshot: { source: "chat:github" } })
.where(eq(heartbeatRuns.id, fixture.runId));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
...(kind === "wrong_destination"
? {
destination: {
endpointId: fixture.endpointId,
conversationId: randomUUID(),
},
}
: {}),
}),
).toBe(false);
if (
[
"gate_changed",
"gate_policy_changed",
"failed_run",
"status_changed",
"revoked_endpoint",
"revoked_principal",
"agent_paused",
"agent_budget_paused",
"agent_terminated",
"agent_pending_approval",
"changed_context",
].includes(kind)
) {
await expect(
issueService(db).addComment(
fixture.issueId,
"Must remain private",
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "allow_chat_run_presentation" },
),
).rejects.toMatchObject({
status: 409,
details: { code: "chat_review_response_presentation_denied" },
});
}
},
);
it("retries a contended review-response comment outside issue and governance locks", async () => {
const fixture = await seedWaitTurn();
await seedPriorCompletionReview(fixture);
const resultJson = await finishReviewResponse(fixture);
let release!: () => void;
let locked!: () => void;
const lockObserved = new Promise<void>((resolve) => {
locked = resolve;
});
const lockRelease = new Promise<void>((resolve) => {
release = resolve;
});
const holding = db.transaction(async (tx) => {
await tx
.select()
.from(chatEndpoints)
.where(eq(chatEndpoints.id, fixture.endpointId))
.for("update");
locked();
await lockRelease;
});
await lockObserved;
let observeBackoff!: () => void;
let releaseBackoff!: () => void;
const backoffObserved = new Promise<void>((resolve) => {
observeBackoff = resolve;
});
const backoffRelease = new Promise<void>((resolve) => {
releaseBackoff = resolve;
});
const originalRetry =
nativeChatReviewPresentation.retryNativeChatReviewPresentation;
const retry = vi
.spyOn(nativeChatReviewPresentation, "retryNativeChatReviewPresentation")
.mockImplementation((attempt) =>
originalRetry(async () => {
try {
return await attempt();
} catch (error) {
if (isExternalChatWaitAuthorizationContention(error)) {
// Observe the actual rollback/backoff boundary, not a wall-clock
// instant that may land inside a subsequent short transaction.
observeBackoff();
await backoffRelease;
}
throw error;
}
}),
);
let finished = false;
const append = issueService(db)
.addComment(
fixture.issueId,
(resultJson.nativeResult as { summary: string }).summary,
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "allow_chat_run_presentation" },
)
.finally(() => {
finished = true;
});
const appendOutcome = append.then(
(comment) => ({ comment }),
(error: unknown) => ({ error }),
);
try {
await backoffObserved;
expect(finished).toBe(false);
await db.transaction(async (tx) => {
await tx
.select()
.from(issues)
.where(eq(issues.id, fixture.issueId))
.for("update", { noWait: true });
await tx
.select()
.from(nativeRunFinalizations)
.where(eq(nativeRunFinalizations.runId, fixture.runId))
.for("update", { noWait: true });
await tx
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.issueId, fixture.issueId))
.for("update", { noWait: true });
});
} finally {
release();
releaseBackoff();
await Promise.allSettled([holding, appendOutcome]);
retry.mockRestore();
}
const outcome = await appendOutcome;
if ("error" in outcome) throw outcome.error;
const { comment } = outcome;
expect(
await db
.select()
.from(chatPublications)
.where(eq(chatPublications.commentId, comment.id)),
).toHaveLength(1);
});
it("keeps earlier queued response files authorized after another wait retains the same review", async () => {
const fixture = await seedWaitTurn();
await seedPriorCompletionReview(fixture);
const resultJson = await finishReviewResponse(fixture);
const [current] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const [accepted] = await db
.select()
.from(nativeRunResults)
.where(eq(nativeRunResults.runId, fixture.runId));
const runId = randomUUID();
const sessionId = randomUUID();
const runnerInstanceId = randomUUID();
await db.insert(heartbeatRuns).values({
id: runId,
companyId: fixture.companyId,
agentId: fixture.agentId,
runtimeMode: "native",
nativeIssueId: fixture.issueId,
nativeSessionId: sessionId,
runnerInstanceId,
status: "running",
completionContractId: current!.completionContractId,
completionContractSha256: current!.completionContractSha256,
contextSnapshot: current!.contextSnapshot,
resultJson: { nativeResult: accepted!.resultJson.result },
startedAt: new Date(),
});
await db
.update(issues)
.set({ executionRunId: runId })
.where(eq(issues.id, fixture.issueId));
const port = new PaperclipControlPlanePort(db, {
companyId: fixture.companyId,
issueId: fixture.issueId,
agentId: fixture.agentId,
runId,
sessionId,
completionContractId: current!.completionContractId!,
completionContractSha256: current!.completionContractSha256!,
sourceInstanceId: runnerInstanceId,
controlPlaneSourceInstanceId: `later-control-${runId}`,
});
await port.openRun({
identity: {
companyId: fixture.companyId,
issueId: fixture.issueId,
agentId: fixture.agentId,
runId,
sessionId,
},
backendKind: "mock",
sourceInstanceId: runnerInstanceId,
});
await port.completeRun({
result: accepted!.resultJson.result as PrpStructuredRunResult,
terminal: accepted!.resultJson.terminal as PrpTerminalState,
});
const laterResult = await finishReviewResponse({ ...fixture, runId });
expect(laterResult.externalChatReviewPresentation).toBeTruthy();
expect(
await authorizeNativeChatReviewPresentation(
db,
{
...fixture,
resultJson,
destination: {
endpointId: fixture.endpointId,
conversationId: fixture.conversationId,
},
},
"read",
),
).toBe(true);
const [firstComment] = await db
.select()
.from(issueComments)
.where(
and(
eq(issueComments.createdByRunId, fixture.runId),
eq(issueComments.authorType, "agent"),
),
);
await db
.delete(chatPublications)
.where(eq(chatPublications.commentId, firstComment!.id));
await db
.delete(issueComments)
.where(eq(issueComments.id, firstComment!.id));
await db
.update(heartbeatRuns)
.set({ resultJson: { keepUnrelatedMetadata: "yes" } })
.where(eq(heartbeatRuns.id, fixture.runId));
await reconcileNativeFinalizations(db, [fixture.runId]);
await reconcileNativeFinalizations(db, [fixture.runId]);
const [repaired] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(repaired!.resultJson).toMatchObject({
keepUnrelatedMetadata: "yes",
externalChatReviewPresentation: resultJson.externalChatReviewPresentation,
});
const comments = await db
.select()
.from(issueComments)
.where(
and(
eq(issueComments.createdByRunId, fixture.runId),
eq(issueComments.authorType, "agent"),
),
);
expect(comments).toHaveLength(1);
expect(
await db
.select()
.from(chatPublications)
.where(eq(chatPublications.commentId, comments[0]!.id)),
).toHaveLength(1);
expect(
await hasMaterializedNativeReviewResponse(db, {
...fixture,
decisionId: String(resultJson.decisionId),
}),
).toBe(true);
expect(
await hasMaterializedNativeReviewResponse(db, {
...fixture,
decisionId: randomUUID(),
}),
).toBe(false);
const updatedAt = repaired!.updatedAt;
await db
.update(issueComments)
.set({ deletedAt: new Date() })
.where(eq(issueComments.id, comments[0]!.id));
await reconcileNativeFinalizations(db, [fixture.runId]);
const [unchanged] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(unchanged!.updatedAt).toEqual(updatedAt);
expect(
await db
.select()
.from(issueComments)
.where(eq(issueComments.createdByRunId, fixture.runId)),
).toEqual([
expect.objectContaining({
id: comments[0]!.id,
deletedAt: expect.any(Date),
}),
]);
});
it.each(["running", "succeeded", "gate_changed", "ownership_held"] as const)(
"repairs the committed presentation crash boundary exactly once: %s",
async (state) => {
const fixture = await seedWaitTurn();
const gate = await seedPriorCompletionReview(fixture);
const selectedComment = await issueService(db).addComment(
fixture.issueId,
"Prepared the original file",
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason: "paperclip_runner_protocol" },
);
const attachment = await issueService(db).createAttachment({
issueId: fixture.issueId,
issueCommentId: selectedComment.id,
provider: "local_disk",
objectKey: `issues/${fixture.issueId}/original.txt`,
contentType: "text/plain",
byteSize: 128,
sha256: "b".repeat(64),
originalFilename: "original.txt",
createdByAgentId: fixture.agentId,
createdByRunId: fixture.runId,
});
// Live finalization commits the grant while heartbeat still owns terminal
// status/presentation. Simulate its process dying before those later steps.
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: false,
});
const [committed] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(
committed!.resultJson?.externalChatReviewPresentation,
).toBeTruthy();
const marker = committed!.resultJson!.externalChatReviewPresentation;
await db
.update(heartbeatRuns)
.set({
status:
state === "succeeded" || state === "gate_changed"
? "succeeded"
: "running",
resultJson: { keepUnrelatedMetadata: "yes" },
...(state === "ownership_held"
? {
nativePhase: "terminal_failure",
errorCode: "native_execution_ownership_unverified",
}
: {}),
})
.where(eq(heartbeatRuns.id, fixture.runId));
if (state === "gate_changed")
await db
.update(issueThreadInteractions)
.set({ status: "accepted" })
.where(eq(issueThreadInteractions.id, gate.id));
const presentationSignals: Array<Record<string, unknown>> = [];
const visibilityChecks: Array<Promise<Array<{ runId: string | null }>>> = [];
const unsubscribe = subscribeAllCompanyLiveEvents((event) => {
if (
event.companyId !== fixture.companyId ||
event.type !== "heartbeat.run.event" ||
event.payload.runId !== fixture.runId ||
event.payload.eventType !== "run.presentation.resolved"
) return;
presentationSignals.push(event.payload);
visibilityChecks.push(
db
.select({ runId: issueComments.createdByRunId })
.from(issueComments)
.where(and(
eq(issueComments.createdByRunId, fixture.runId),
eq(issueComments.authorType, "agent"),
ne(issueComments.id, selectedComment.id),
)),
);
});
try {
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
} finally {
unsubscribe();
}
const [repaired] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(repaired!.resultJson?.keepUnrelatedMetadata).toBe("yes");
if (state === "ownership_held") {
expect(repaired).toMatchObject({
status: "running",
nativePhase: "terminal_failure",
errorCode: "native_execution_ownership_unverified",
});
} else {
expect(repaired!.resultJson?.externalChatReviewPresentation).toEqual(
marker,
);
expect(repaired!.status).toBe("succeeded");
}
const publications = await db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId));
if (state === "gate_changed" || state === "ownership_held")
expect(publications).toEqual([]);
else {
expect(publications).toHaveLength(2);
expect(publications).toEqual(
expect.arrayContaining([
expect.objectContaining({
idempotencyKey: `attachment:${attachment.id}:${fixture.endpointId}`,
}),
]),
);
}
if (state === "gate_changed" || state === "ownership_held") {
expect(presentationSignals).toEqual([]);
expect(visibilityChecks).toEqual([]);
} else {
expect(presentationSignals).toEqual([{
runId: fixture.runId,
agentId: fixture.agentId,
issueId: fixture.issueId,
eventType: "run.presentation.resolved",
}]);
await expect(Promise.all(visibilityChecks)).resolves.toEqual([
[{ runId: fixture.runId }],
]);
}
expect(
await db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).toHaveLength(2);
expect(
await db
.select()
.from(issueThreadInteractions)
.where(eq(issueThreadInteractions.issueId, fixture.issueId)),
).toHaveLength(1);
expect(
await db
.select()
.from(agentWakeupRequests)
.where(eq(agentWakeupRequests.companyId, fixture.companyId)),
).toEqual([]);
},
);
it("does not overwrite an ownership hold racing the committed replay projection", async () => {
const fixture = await seedWaitTurn();
await seedPriorCompletionReview(fixture);
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: false,
});
// Arrange the retained execution lock explicitly: the earlier review
// status projection is allowed to have released the normal run lock.
await db
.update(issues)
.set({ executionRunId: fixture.runId })
.where(eq(issues.id, fixture.issueId));
let locked!: () => void;
let release!: () => void;
const lockObserved = new Promise<void>((resolve) => {
locked = resolve;
});
const lockRelease = new Promise<void>((resolve) => {
release = resolve;
});
const holder = db.transaction(async (tx) => {
await tx
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId))
.for("update");
locked();
await lockRelease;
await tx
.update(heartbeatRuns)
.set({
status: "running",
nativePhase: "terminal_failure",
errorCode: "native_execution_ownership_unverified",
})
.where(eq(heartbeatRuns.id, fixture.runId));
});
await lockObserved;
const replay = finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
try {
await vi.waitFor(
async () => {
let coordinatorOwned = false;
try {
await db.transaction((tx) =>
tx
.select()
.from(nativeRunFinalizations)
.where(eq(nativeRunFinalizations.runId, fixture.runId))
.for("update", { noWait: true }),
);
} catch (error) {
coordinatorOwned = isExternalChatWaitAuthorizationContention(error);
}
expect(coordinatorOwned).toBe(true);
},
{ interval: 5, timeout: 1_000 },
);
} finally {
release();
await holder;
}
await replay;
const [run] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(run).toMatchObject({
status: "running",
nativePhase: "terminal_failure",
errorCode: "native_execution_ownership_unverified",
});
const [issue] = await db
.select()
.from(issues)
.where(eq(issues.id, fixture.issueId));
expect(issue!.executionRunId).toBe(fixture.runId);
expect(
await db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId)),
).toEqual([]);
});
it("keeps presentation authority when heartbeat appends only adapter runtime-service display metadata", async () => {
const fixture = await seedWaitTurn();
await seedPriorCompletionReview(fixture);
const resultJson = await finishReviewResponse(fixture);
const [run] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const context = {
...run!.contextSnapshot,
paperclipRuntimeServices: [
{ name: "preview", url: "http://127.0.0.1:9000" },
],
paperclipRuntimePrimaryUrl: "http://127.0.0.1:9000",
};
await db
.update(heartbeatRuns)
.set({ contextSnapshot: context })
.where(eq(heartbeatRuns.id, fixture.runId));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(true);
await db
.update(heartbeatRuns)
.set({ contextSnapshot: { ...context, wakeCommentIds: [randomUUID()] } })
.where(eq(heartbeatRuns.id, fixture.runId));
expect(
await authorizeNativeChatReviewPresentation(db, {
...fixture,
resultJson,
}),
).toBe(false);
});
it("does not rewrite an already-materialized latest response during real reconciliation", async () => {
const fixture = await seedWaitTurn();
await seedPriorCompletionReview(fixture);
await finishReviewResponse(fixture);
await db
.insert(workspaceOperations)
.values({
companyId: fixture.companyId,
issueId: fixture.issueId,
heartbeatRunId: fixture.runId,
phase: "workspace_finalize",
status: "succeeded",
finishedAt: new Date(),
});
const [before] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
await reconcileNativeFinalizations(db, [fixture.runId]);
await reconcileNativeFinalizations(db, [fixture.runId]);
const [after] = await db
.select()
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
expect(after!.updatedAt).toEqual(before!.updatedAt);
expect(after!.resultJson).toEqual(before!.resultJson);
expect(
await db
.select()
.from(chatPublications)
.where(eq(chatPublications.issueId, fixture.issueId)),
).toHaveLength(1);
});
it.each([
["Slack fresh setup", "slack", 1],
["GitHub fresh setup", "github", 1],
["Discord fresh setup", "discord", 1],
["Teams fresh setup", "microsoft-teams", 1],
["Telegram fresh setup", "telegram", 1],
["GitHub reconnect setup", "github", 8],
])(
"authorizes an exact response wait during the current %s test window",
async (_label, provider, generation) => {
const fixture = await seedWaitTurn(
provider as Parameters<typeof seedWaitTurn>[0],
);
if (provider === "github") {
// Only upgraded GitHub bots keep the selected final internal. Legacy
// connection fixtures retain their automatic presentation behavior.
await db.insert(chatGitHubConfigurations).values({
companyId: fixture.companyId,
endpointId: fixture.endpointId,
configuration: {
version: 1,
toolsEnabled: true,
responsibleUserId: fixture.userId,
memberAccess: "all_linked",
people: [],
defaults: defaultGitHubReviewPolicy(),
repositories: {},
},
updatedByUserId: fixture.userId,
});
}
await placeWaitTurnInSetupTest(fixture, { generation });
await expect(
resolveExternalChatResponseWaitAuthorization({
db,
binding: fixture,
}),
).resolves.toBe("authorized");
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
await expect(
db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({
reasonCode: "external_chat_response_waiting",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
const [finalizedRun] = await db
.select({ resultJson: heartbeatRuns.resultJson })
.from(heartbeatRuns)
.where(eq(heartbeatRuns.id, fixture.runId));
const authorizationReason = provider === "github"
? "internal_agent_write" : "allow_chat_run_presentation";
await expect(
resolveChatRunPresentationAuthorizationReason(db, fixture),
).resolves.toBe(authorizationReason);
const response = resolveHeartbeatRunResponse({
resultJson: finalizedRun!.resultJson,
preferFinalResponseOverExistingComment: true,
externalChatResponseWakeSummaryAuthorized: true,
});
expect(response).toEqual(
expect.objectContaining({
text: "The requested photo is prepared. I will wait for your next message.",
decision: expect.objectContaining({ commentAction: "create" }),
}),
);
const comment = await issueService(db).addComment(
fixture.issueId,
response.text!,
{ agentId: fixture.agentId, runId: fixture.runId },
{ authorizationReason },
);
await expect(
db
.select()
.from(chatPublications)
.where(eq(chatPublications.commentId, comment.id)),
).resolves.toEqual(provider === "github" ? [] : [
expect.objectContaining({
conversationId: fixture.conversationId,
endpointId: fixture.endpointId,
state: "pending",
}),
]);
},
);
it("rejects stale or revoked setup-test provenance without weakening active policy", async () => {
type Fixture = Awaited<ReturnType<typeof seedWaitTurn>>;
type Times = Awaited<ReturnType<typeof placeWaitTurnInSetupTest>>;
const expectRevoked = async (
label: string,
mutate: (fixture: Fixture, times: Times) => Promise<void>,
) => {
const fixture = await seedWaitTurn("github");
const times = await placeWaitTurnInSetupTest(fixture, { generation: 4 });
await mutate(fixture, times);
await expect(
resolveExternalChatResponseWaitAuthorization({
db,
binding: fixture,
}),
label,
).resolves.toBe("revoked");
};
await expectRevoked("received before the current test", async (fixture, times) => {
await db
.update(chatDeliveries)
.set({ receivedAt: new Date(times.testStartedAt.getTime() - 1) })
.where(eq(chatDeliveries.id, fixture.deliveryId));
});
await expectRevoked("processed before the current test", async (fixture, times) => {
await db
.update(chatDeliveries)
.set({ processedAt: new Date(times.testStartedAt.getTime() - 1) })
.where(eq(chatDeliveries.id, fixture.deliveryId));
});
await expectRevoked("prior runtime generation", async (fixture) => {
await db
.update(chatDeliveries)
.set({
normalizedEvent: {
runtimeContext: {
generation: 3,
credentialFingerprint: "a".repeat(64),
},
},
})
.where(eq(chatDeliveries.id, fixture.deliveryId));
});
await expectRevoked("malformed runtime generation", async (fixture) => {
await db.execute(sql`
update chat_endpoints
set setup = jsonb_set(setup, '{runtimeGeneration}', '"4"'::jsonb)
where id = ${fixture.endpointId}
`);
});
await expectRevoked("invalid runtime fingerprint", async (fixture) => {
await db
.update(chatDeliveries)
.set({
normalizedEvent: {
runtimeContext: {
generation: 4,
credentialFingerprint: "not-a-runtime-fingerprint",
},
},
})
.where(eq(chatDeliveries.id, fixture.deliveryId));
});
await expectRevoked("missing runtime fence", async (fixture) => {
await db
.update(chatDeliveries)
.set({ normalizedEvent: {} })
.where(eq(chatDeliveries.id, fixture.deliveryId));
});
await expectRevoked("invalid test timestamp", async (fixture) => {
await db
.update(chatEndpoints)
.set({
setup: {
step: "test",
testStartedAt: "not-a-timestamp",
runtimeGeneration: 4,
} as (typeof chatEndpoints.$inferSelect)["setup"] & {
runtimeGeneration: number;
},
})
.where(eq(chatEndpoints.id, fixture.endpointId));
});
await expectRevoked("wrong setup step", async (fixture, times) => {
await db
.update(chatEndpoints)
.set({
setup: {
step: "provider_setup",
testStartedAt: times.testStartedAt.toISOString(),
runtimeGeneration: 4,
} as (typeof chatEndpoints.$inferSelect)["setup"] & {
runtimeGeneration: number;
},
})
.where(eq(chatEndpoints.id, fixture.endpointId));
});
for (const status of ["paused", "revoked"] as const) {
await expectRevoked(`${status} endpoint`, async (fixture) => {
await db
.update(chatEndpoints)
.set({ status })
.where(eq(chatEndpoints.id, fixture.endpointId));
});
}
await expectRevoked("expired principal link", async (fixture) => {
await db
.update(chatIdentityLinks)
.set({ status: "expired", updatedAt: new Date() })
.where(eq(chatIdentityLinks.principalId, fixture.principalId));
});
await expectRevoked("disabled current resource", async (fixture) => {
await db
.update(chatEndpointResources)
.set({ enabled: false, updatedAt: new Date() })
.where(eq(chatEndpointResources.id, fixture.resourceId));
});
});
it("loses setup-test response authority when reconnect rotates the window before commit", async () => {
const fixture = await seedWaitTurn("github");
await placeWaitTurnInSetupTest(fixture, { generation: 4 });
let releaseEndpoint!: () => void;
let endpointLocked!: () => void;
const endpointRelease = new Promise<void>((resolve) => {
releaseEndpoint = resolve;
});
const endpointLockObserved = new Promise<void>((resolve) => {
endpointLocked = resolve;
});
const endpointHolder = db.transaction(async (tx) => {
await tx
.select({ id: chatEndpoints.id })
.from(chatEndpoints)
.where(eq(chatEndpoints.id, fixture.endpointId))
.for("update");
await tx
.update(chatEndpoints)
.set({
setup: {
step: "test",
testStartedAt: new Date().toISOString(),
runtimeGeneration: 5,
} as (typeof chatEndpoints.$inferSelect)["setup"] & {
runtimeGeneration: number;
},
updatedAt: new Date(),
})
.where(eq(chatEndpoints.id, fixture.endpointId));
endpointLocked();
await endpointRelease;
});
await endpointLockObserved;
const finalization = finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
const joined = Promise.allSettled([endpointHolder, finalization]);
let waitError: unknown = null;
try {
await vi.waitFor(
async () => {
const [coordinator] = await db
.select({ phase: nativeRunFinalizations.phase })
.from(nativeRunFinalizations)
.where(eq(nativeRunFinalizations.runId, fixture.runId));
expect(coordinator?.phase).toBe("arbitrating");
},
{ timeout: 1_000, interval: 5 },
);
} catch (error) {
waitError = error;
} finally {
releaseEndpoint();
}
const [holderOutcome, finalizationOutcome] = await joined;
if (waitError) throw waitError;
if (holderOutcome.status === "rejected") throw holderOutcome.reason;
if (finalizationOutcome.status === "rejected") {
throw finalizationOutcome.reason;
}
await expect(
db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({
reasonCode: "external_chat_response_wait_authorization_lost",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
});
it("retains setup-test response authority when the endpoint activates before commit", async () => {
const fixture = await seedWaitTurn("github");
await placeWaitTurnInSetupTest(fixture, { generation: 4 });
let releaseEndpoint!: () => void;
let endpointLocked!: () => void;
const endpointRelease = new Promise<void>((resolve) => {
releaseEndpoint = resolve;
});
const endpointLockObserved = new Promise<void>((resolve) => {
endpointLocked = resolve;
});
const endpointHolder = db.transaction(async (tx) => {
await tx
.select({ id: chatEndpoints.id })
.from(chatEndpoints)
.where(eq(chatEndpoints.id, fixture.endpointId))
.for("update");
await tx
.update(chatEndpoints)
.set({
status: "active",
setup: { step: "complete", testStartedAt: null },
updatedAt: new Date(),
})
.where(eq(chatEndpoints.id, fixture.endpointId));
endpointLocked();
await endpointRelease;
});
await endpointLockObserved;
const finalization = finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
const joined = Promise.allSettled([endpointHolder, finalization]);
let waitError: unknown = null;
try {
await vi.waitFor(
async () => {
const [coordinator] = await db
.select({ phase: nativeRunFinalizations.phase })
.from(nativeRunFinalizations)
.where(eq(nativeRunFinalizations.runId, fixture.runId));
expect(coordinator?.phase).toBe("arbitrating");
},
{ timeout: 1_000, interval: 5 },
);
} catch (error) {
waitError = error;
} finally {
releaseEndpoint();
}
const [holderOutcome, finalizationOutcome] = await joined;
if (waitError) throw waitError;
if (holderOutcome.status === "rejected") throw holderOutcome.reason;
if (finalizationOutcome.status === "rejected") {
throw finalizationOutcome.reason;
}
await expect(
db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({
reasonCode: "external_chat_response_waiting",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
});
it("parks a verified external-chat response wait without scheduling work", async () => {
const fixture = await seedWaitTurn();
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
await expect(
db.select().from(issues).where(eq(issues.id, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({ status: "in_progress", statusVersion: 1 }),
]);
await expect(
db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({
toStatus: "in_progress",
reasonCode: "external_chat_response_waiting",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
await expect(
db
.select()
.from(agentWakeupRequests)
.where(
and(
eq(agentWakeupRequests.companyId, fixture.companyId),
eq(agentWakeupRequests.agentId, fixture.agentId),
),
),
).resolves.toEqual([]);
await expect(
db
.select()
.from(chatConversations)
.where(eq(chatConversations.id, fixture.conversationId)),
).resolves.toEqual([
expect.objectContaining({
issueId: fixture.issueId,
sessionGeneration: 1,
state: "active",
}),
]);
});
it.each([
[3, "yielded", "in_review", "response_wake", false],
[5, "yielded", "in_review", "response_wake", false],
[5, "done", "done", null, false],
[3, "yielded", "in_review", "monitor", false],
[3, "yielded", "in_review", "response_wake", true],
[3, "yielded", "in_review", "response_wake", false, "tree_resume"],
[3, "yielded", "in_review", "response_wake", false, "continued_review"],
[3, "yielded", "in_review", "monitor", false, "continued_review"],
[3, "yielded", "in_progress", "response_wake", false, "new_head"],
] as const)("settles a published %s/5 review reporting %s in %s with %s, separate approval: %s, resume: %s", async (score, disposition, status, continuationKind, separateApproval, resumeMode?: "tree_resume" | "continued_review" | "new_head") => {
const resumed = resumeMode !== undefined;
const savedPublication = resumeMode === "continued_review" || resumeMode === "new_head";
const unreviewedHead = resumeMode === "new_head";
const fixture = await seedWaitTurn("github", [], {
reportedWorkDisposition: disposition,
summary: disposition === "done" ? "Review is 5/5 and required checks passed." : "Review published; waiting for the next PR update.",
...(disposition === "done" ? { continuation: undefined } : {}),
...(continuationKind === "monitor" ? { continuation: { kind: "monitor", summary: "Wait for the saved check.", idempotencyKey: "review-monitor" } } : {}),
});
await db.update(issues).set({ title: "Review the pull request", originKind: "chat_channel",
createdByUserId: fixture.userId, responsibleUserId: fixture.userId }).where(eq(issues.id, fixture.issueId));
const oldRequest = score === 3 && !resumed ? await seedPriorCompletionReview(fixture) : null;
if (continuationKind === "monitor") {
await new PaperclipRunnerToolAuthority(db, fixture).execute({ tool: "set_task_monitor", callId: randomUUID(),
arguments: { idempotencyKey: "review-monitor", monitor: { nextCheckAt: new Date(Date.now() + 60_000).toISOString(), notes: "Check PR status." } } });
}
const [review] = await db.insert(chatGitHubReviews).values({
companyId: fixture.companyId, endpointId: fixture.endpointId, issueId: fixture.issueId,
runId: fixture.runId, repositoryId: "1", repository: "paperclip/test-repository", pullNumber: 1,
headSha: "a".repeat(40), deliveryId: fixture.deliveryId, configurationRevision: 1,
policySnapshot: defaultGitHubReviewPolicy(), event: {
event: "opened", deliveryId: fixture.deliveryId, repositoryId: "1", repository: "paperclip/test-repository",
pullNumber: 1, title: "Review the pull request", body: "", baseSha: "b".repeat(40), headSha: "a".repeat(40),
baseBranch: "master", author: { id: "1", login: "author", isBot: false }, sender: { id: "1", login: "author" },
draft: false, labels: [],
},
state: "completed", assessment: { complete: true, score, reviewedCommit: "a".repeat(40),
summary: "Review published", rationale: "Review complete", findings: [],
coverage: { reviewedPaths: [], omittedPaths: [], limitations: [] },
},
checkId: "123",
}).returning();
await expect(hasPublishedGitHubReview(db, fixture)).resolves.toBe(false);
// A completed but unpublished head must not hide another published head from this run.
const [published] = await db.insert(chatGitHubReviews).values({ ...review!, id: randomUUID(), deliveryId: randomUUID(), headSha: "c".repeat(40), assessment: { ...review!.assessment!, reviewedCommit: "c".repeat(40) }, createdAt: new Date(Date.now() + 10) }).returning();
await db.insert(chatActions).values({
companyId: fixture.companyId, endpointId: fixture.endpointId, kind: "github_review_publication",
providerActionId: `github_publication:${fixture.issueId}:assessment:${createHash("sha256").update(JSON.stringify(published!.id)).digest("hex")}`,
status: "processed", payload: { operation: "assessment", reviewId: published!.id },
result: { reviewId: published!.id, checkUrl: "https://github.com/paperclip/test-repository/runs/123" },
});
if (oldRequest) {
const [older] = await db.insert(chatGitHubReviews).values({
...review!, id: randomUUID(), runId: oldRequest.sourceRunId, deliveryId: randomUUID(),
}).returning();
await db.insert(chatActions).values({ companyId: fixture.companyId, endpointId: fixture.endpointId,
kind: "github_review_publication", status: "processed",
providerActionId: `github_publication:${fixture.issueId}:assessment:${createHash("sha256").update(JSON.stringify(older!.id)).digest("hex")}`,
});
}
let inheritedOriginalRunId: string | null = null;
if (resumed) {
await db.update(chatEndpointResources).set({ metadata: { providerRepositoryId: "1" } }).where(eq(chatEndpointResources.id, fixture.resourceId));
await db.update(chatConversations).set({ externalThreadId: "github:paperclip/test-repository:1" }).where(eq(chatConversations.id, fixture.conversationId));
// A tree resume inherits its requester through the accepted identity
// ledger, while its immediate wake correctly has no chat envelope.
const [run] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
const originalRunId = randomUUID();
inheritedOriginalRunId = originalRunId;
await db.insert(heartbeatRuns).values({ ...run!, id: originalRunId, activeIdentityContextId: null, status: "succeeded", finishedAt: new Date() });
const [origin] = await db.insert(runIdentityContexts).values({
companyId: fixture.companyId, runId: originalRunId, revision: 1,
messageId: fixture.commentId, responsibleUserId: fixture.userId,
cause: "chat_message", correlationId: "dispatch", status: "accepted",
}).returning();
const [continuation] = await db.insert(runIdentityContexts).values({
companyId: fixture.companyId, runId: fixture.runId, revision: 1,
parentContextId: origin!.id, responsibleUserId: fixture.userId,
cause: "issue_tree_resumed", correlationId: "dispatch", status: "accepted",
}).returning();
await db.update(heartbeatRuns).set({ activeIdentityContextId: continuation!.id,
responsibleUserId: fixture.userId, contextSnapshot: {
source: "issue.tree_resume", wakeReason: "issue_tree_resumed",
issueId: fixture.issueId, paperclipHarnessCheckedOut: true,
paperclipWake: { checkedOutByHarness: true, issue: { id: fixture.issueId }, commentIds: [] },
},
}).where(eq(heartbeatRuns.id, fixture.runId));
}
await expect(hasPublishedGitHubReview(db, fixture)).resolves.toBe(true);
await expect(hasPublishedGitHubReview(db, { ...fixture, runId: randomUUID() })).resolves.toBe(false);
if (resumed) {
const [inheritedRun] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
await expect(readInheritedGitHubReviewWait(db, inheritedRun!)).resolves.toMatchObject({ kind: "published" });
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.toBe("authorized");
for (const binding of [
{ ...fixture, companyId: randomUUID() }, { ...fixture, issueId: randomUUID() },
{ ...fixture, agentId: randomUUID() },
]) await expect(resolveExternalChatResponseWaitAuthorization({ db, binding })).resolves.not.toBe("authorized");
await db.update(chatActions).set({ status: "pending" }).where(eq(chatActions.companyId, fixture.companyId));
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.not.toBe("authorized");
await db.update(chatActions).set({ status: "processed" }).where(eq(chatActions.companyId, fixture.companyId));
await db.update(chatEndpointResources).set({ enabled: false }).where(eq(chatEndpointResources.endpointId, fixture.endpointId));
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.toBe("revoked");
await db.update(chatEndpointResources).set({ enabled: true }).where(eq(chatEndpointResources.endpointId, fixture.endpointId));
await db.update(heartbeatRuns).set({ responsibleUserId: "different-requester" }).where(eq(heartbeatRuns.id, fixture.runId));
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.not.toBe("authorized");
await db.update(heartbeatRuns).set({ responsibleUserId: fixture.userId }).where(eq(heartbeatRuns.id, fixture.runId));
await db.update(chatConversations).set({ externalThreadId: "github:paperclip/other-repository:1" }).where(eq(chatConversations.id, fixture.conversationId));
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.not.toBe("authorized");
await db.update(chatConversations).set({ externalThreadId: "github:paperclip/test-repository:1" }).where(eq(chatConversations.id, fixture.conversationId));
await db.update(chatEndpointResources).set({ metadata: { providerRepositoryId: "2" } }).where(eq(chatEndpointResources.id, fixture.resourceId));
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.not.toBe("authorized");
await db.update(chatEndpointResources).set({ metadata: { providerRepositoryId: "1" } }).where(eq(chatEndpointResources.id, fixture.resourceId));
const [newHead] = await db.insert(chatGitHubReviews).values({ ...review!, id: randomUUID(),
deliveryId: randomUUID(), runId: null, headSha: "d".repeat(40), state: "queued", createdAt: new Date(Date.now() + 1000),
}).returning();
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.not.toBe("authorized");
await db.delete(chatGitHubReviews).where(eq(chatGitHubReviews.id, newHead!.id));
if (savedPublication) {
await db.update(chatGitHubReviews).set({ runId: inheritedOriginalRunId }).where(eq(chatGitHubReviews.issueId, fixture.issueId));
await expect(hasPublishedGitHubReview(db, fixture)).resolves.toBe(false);
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.toBe("authorized");
}
}
if (unreviewedHead) {
await db.insert(chatGitHubReviews).values({ ...review!, id: randomUUID(), runId: null,
deliveryId: randomUUID(), headSha: "e".repeat(40), state: "queued", createdAt: new Date(Date.now() + 1000),
});
await expect(resolveExternalChatResponseWaitAuthorization({ db, binding: fixture })).resolves.toBe("revoked");
}
const [stored] = await db.select().from(nativeRunResults).where(eq(nativeRunResults.runId, fixture.runId));
await expect(nativeCompletionFeedback(db, fixture.runId, stored!.resultJson.result as PrpStructuredRunResult)).resolves.toContain("accepted");
if (oldRequest) {
const pending = async () => (await db.select().from(issueThreadInteractions).where(eq(issueThreadInteractions.id, oldRequest.id)))[0]!.status;
expect(await pending()).toBe("pending");
const reported = stored!.resultJson.result as PrpStructuredRunResult;
await expect(nativeCompletionFeedback(db, fixture.runId, { ...reported,
completionClaim: { ...reported.completionClaim, contractRevision: "stale" } })).rejects.toThrow("Stale completionClaim");
expect(await pending()).toBe("pending");
if (separateApproval) {
const [approval] = await db.insert(approvals).values({ companyId: fixture.companyId, type: "hire_agent", status: "pending", payload: {} }).returning();
await db.insert(issueApprovals).values({ companyId: fixture.companyId, issueId: fixture.issueId, approvalId: approval!.id });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect(await pending()).toBe("pending");
expect((await issueService(db).list(fixture.companyId, { touchedByUserId: fixture.userId })).map(issue => issue.id)).toContain(fixture.issueId);
return;
}
expect(await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true,
failpoint: "status_projection" })).toMatchObject({ phase: "retryable_failure" });
expect(await pending()).toBe("pending");
}
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
const all = await issueService(db).list(fixture.companyId);
expect(all.find(issue => issue.id === fixture.issueId)).toMatchObject({ status,
...(status === "in_review" ? { reviewAttention: { state: "covered", paths: [expect.objectContaining({ kind: continuationKind === "monitor" ? "monitor" : "external_response" })] } } : {}),
});
expect(await issueService(db).list(fixture.companyId, { touchedByUserId: fixture.userId })).toEqual([]);
expect(await db.select().from(issueThreadInteractions).where(and(
eq(issueThreadInteractions.issueId, fixture.issueId), eq(issueThreadInteractions.status, "pending"),
))).toEqual([]);
if (oldRequest) expect(await db.select().from(issueThreadInteractions).where(eq(issueThreadInteractions.id, oldRequest.id)))
.toEqual([expect.objectContaining({ status: "cancelled", result: expect.objectContaining({ outcome: "withdrawn" }) })]);
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toEqual([]);
if (resumed) {
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toEqual([]);
expect(await db.select().from(chatActions).where(and(eq(chatActions.companyId, fixture.companyId), eq(chatActions.kind, "github_review_publication")))).toHaveLength(1);
if (unreviewedHead) {
expect((await db.select().from(statusDecisions).where(eq(statusDecisions.runId, fixture.runId)))[0])
.toMatchObject({ reasonCode: "external_chat_response_wait_authorization_lost" });
}
}
if (status === "in_review" && continuationKind === "response_wake") {
const attention = async () => (await issueService(db).listReviewAttention(fixture.companyId, [{ id: fixture.issueId, companyId: fixture.companyId, status }])).get(fixture.issueId)?.state;
await db.update(chatEndpointResources).set({ enabled: false }).where(eq(chatEndpointResources.endpointId, fixture.endpointId));
expect(await attention()).toBe("stalled");
await db.update(chatEndpointResources).set({ enabled: true }).where(eq(chatEndpointResources.endpointId, fixture.endpointId));
await db.update(companyMemberships).set({ status: "suspended" }).where(eq(companyMemberships.companyId, fixture.companyId));
expect(await attention()).toBe("stalled");
await db.update(companyMemberships).set({ status: "active" }).where(eq(companyMemberships.companyId, fixture.companyId));
await db.update(chatEndpoints).set({ status: "revoked" }).where(eq(chatEndpoints.id, fixture.endpointId));
expect((await issueService(db).listReviewAttention(fixture.companyId, [{
id: fixture.issueId, companyId: fixture.companyId, status,
}])).get(fixture.issueId)?.state).toBe("stalled");
}
});
async function seedPublishedReviewCompletionTurn(contractOverride?: Parameters<typeof seedWaitTurn>[4]) {
const fixture = await seedWaitTurn("github", [], {}, true, contractOverride);
await db.update(chatEndpointResources).set({ metadata: { providerRepositoryId: "1" } })
.where(eq(chatEndpointResources.id, fixture.resourceId));
await db.update(chatConversations).set({ externalThreadId: "github:paperclip/test-repository:1" })
.where(eq(chatConversations.id, fixture.conversationId));
const [identity] = await db.insert(runIdentityContexts).values({ companyId: fixture.companyId,
runId: fixture.runId, revision: 1, messageId: fixture.commentId, responsibleUserId: fixture.userId,
cause: "chat_message", correlationId: "dispatch", status: "accepted" }).returning();
await db.update(heartbeatRuns).set({ activeIdentityContextId: identity!.id, responsibleUserId: fixture.userId })
.where(eq(heartbeatRuns.id, fixture.runId));
const [review] = await db.insert(chatGitHubReviews).values({ companyId: fixture.companyId,
endpointId: fixture.endpointId, issueId: fixture.issueId, runId: fixture.runId,
repositoryId: "1", repository: "paperclip/test-repository", pullNumber: 1, headSha: "a".repeat(40),
deliveryId: fixture.deliveryId, configurationRevision: 1, policySnapshot: defaultGitHubReviewPolicy(),
event: { event: "opened", deliveryId: fixture.deliveryId, repositoryId: "1", repository: "paperclip/test-repository",
pullNumber: 1, title: "Review", body: "", baseSha: "b".repeat(40), headSha: "a".repeat(40),
baseBranch: "master", author: { id: "1", login: "author", isBot: false }, sender: { id: "1", login: "author" }, draft: false, labels: [] },
state: "completed", assessment: { complete: true, score: 3, reviewedCommit: "a".repeat(40),
summary: "Quantity is ignored", rationale: "Review complete", findings: [],
coverage: { reviewedPaths: [], omittedPaths: [], limitations: [] } }, checkId: "123" }).returning();
await db.insert(chatActions).values({ companyId: fixture.companyId, endpointId: fixture.endpointId,
kind: "github_review_publication", status: "processed", payload: { operation: "assessment", reviewId: review!.id },
providerActionId: `github_publication:${fixture.issueId}:assessment:${createHash("sha256").update(JSON.stringify(review!.id)).digest("hex")}` });
return { ...fixture, review: review! };
}
it.each([false, true])("binds a low-score Retry wait to its accepted newer source after a predecessor failed before identity: %s", async acceptedSource => {
const objective = "Review the current head. A lower score stays In review with a passive response wake; only a merged PR or 5/5 with passing checks is Done.";
const fixture = await seedPublishedReviewCompletionTurn({ objective,
criteria: [{ id: "response", requirement: objective }] });
await db.update(chatGitHubReviews).set({ policySnapshot: { ...fixture.review.policySnapshot, ratingThreshold: 5 },
assessment: { ...fixture.review.assessment!, score: 2 } }).where(eq(chatGitHubReviews.id, fixture.review.id));
const previousRunId = randomUUID();
await db.insert(heartbeatRuns).values({ id: previousRunId, companyId: fixture.companyId, agentId: fixture.agentId,
issueId: fixture.issueId, status: "failed", errorCode: "chat_failed_run_retry_not_authorized",
finishedAt: new Date(), contextSnapshot: { issueId: fixture.issueId, source: "chat:github" } });
const [run] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
await db.update(heartbeatRuns).set({ activeIdentityContextId: null, retryOfRunId: previousRunId,
contextSnapshot: { ...run!.contextSnapshot, retryOfRunId: previousRunId, wakeReason: "retry_failed_run" } })
.where(eq(heartbeatRuns.id, fixture.runId));
await db.delete(runIdentityContexts).where(eq(runIdentityContexts.id, run!.activeIdentityContextId!));
await initializeRunIdentity(db, { companyId: fixture.companyId, issueId: fixture.issueId,
runId: fixture.runId, responsibleUserId: fixture.userId, parentRunId: previousRunId,
messageIds: acceptedSource ? [fixture.commentId] : [], cause: "retry_failed_run" });
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const read = () => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} });
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
const report = { ...fixture.result, continuation: undefined, completionClaim: {
...fixture.result.completionClaim, objectiveSatisfied: false,
criteria: [{ criterionId: "response", status: "unknown" as const, evidenceRefs: [] }],
remainingWork: [{ description: "Author must fix the quantity defect before this review can reach the passing score.", blocksCompletion: true }],
} };
if (!acceptedSource) {
await expect(resolveCompletionContinuation(report, fixture.runId, read)).rejects.toThrow("no eligible saved monitor");
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0].status).not.toBe("done");
return;
}
const resolved = await resolveCompletionContinuation(report, fixture.runId, read) as PrpStructuredRunResult;
expect(resolved).toMatchObject({ reportedWorkDisposition: "yielded", continuation: { kind: "response_wake" } });
await expect(nativeCompletionFeedback(db, fixture.runId, resolved)).resolves.toContain("accepted");
await fixture.port.completeRun({ result: resolved, terminal: fixture.terminal, callerResultId: `retry-wait-${fixture.runId}` });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0]).toMatchObject({ status: "in_review" });
expect((await db.select().from(statusDecisions).where(eq(statusDecisions.runId, fixture.runId)))[0])
.toMatchObject({ reasonCode: "external_chat_response_waiting", toStatus: "in_review", decisionJson: { effects: [] } });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
expect(await db.select().from(chatActions).where(and(eq(chatActions.companyId, fixture.companyId),
eq(chatActions.kind, "github_review_publication")))).toHaveLength(1);
});
async function seedInheritedReviewCompletionTurn(resume: boolean) {
const fixture = await seedPublishedReviewCompletionTurn();
const [run] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
const predecessorId = randomUUID();
await db.insert(heartbeatRuns).values({ ...run!, id: predecessorId, activeIdentityContextId: null,
status: "succeeded", finishedAt: new Date() });
await db.update(chatGitHubReviews).set({ runId: predecessorId }).where(eq(chatGitHubReviews.id, fixture.review.id));
if (resume) {
const [origin] = await db.insert(runIdentityContexts).values({ companyId: fixture.companyId,
runId: predecessorId, revision: 1, messageId: fixture.commentId, responsibleUserId: fixture.userId,
cause: "chat_message", correlationId: "dispatch", status: "accepted" }).returning();
await db.update(runIdentityContexts).set({ messageId: null, parentContextId: origin!.id,
cause: "issue_tree_resumed" }).where(eq(runIdentityContexts.id, run!.activeIdentityContextId!));
await db.update(heartbeatRuns).set({ contextSnapshot: { source: "issue.tree_resume", wakeReason: "issue_tree_resumed",
issueId: fixture.issueId, paperclipHarnessCheckedOut: true,
paperclipWake: { checkedOutByHarness: true, issue: { id: fixture.issueId }, commentIds: [] } },
}).where(eq(heartbeatRuns.id, fixture.runId));
}
return { ...fixture, predecessorId };
}
it.each([undefined, null, "malformed-first"] as const)("derives an omitted inherited GitHub review wait after normal tree resume: %s", async absent => {
const fixture = await seedInheritedReviewCompletionTurn(true);
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const read = () => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} });
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
const report = { ...fixture.result, continuation: absent === "malformed-first" ? "response_wake" : absent };
if (absent === "malformed-first") {
expect(await resolveCompletionContinuation(report, fixture.runId, read)).toEqual(report);
expect(validatePrpStructuredRunResult(report).ok).toBe(false);
delete report.continuation;
}
const resolved = await resolveCompletionContinuation(report, fixture.runId, read) as PrpStructuredRunResult;
expect(resolved).toMatchObject({ reportedWorkDisposition: "yielded", continuation: { kind: "response_wake" } });
await expect(nativeCompletionFeedback(db, fixture.runId, resolved)).resolves.toContain("accepted");
await fixture.port.completeRun({ result: resolved, terminal: fixture.terminal, callerResultId: `wait-result-${fixture.runId}` });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0]).toMatchObject({ status: "in_review" });
expect((await db.select().from(statusDecisions).where(eq(statusDecisions.runId, fixture.runId)))[0])
.toMatchObject({ reasonCode: "external_chat_response_waiting", decisionJson: { effects: [] } });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
expect(await db.select().from(chatActions).where(and(eq(chatActions.companyId, fixture.companyId),
eq(chatActions.kind, "github_review_publication")))).toHaveLength(1);
});
async function seedCurrentAuthorReply(fixture: Awaited<ReturnType<typeof seedInheritedReviewCompletionTurn>>) {
const [reply] = await db.insert(chatActions).values({ companyId: fixture.companyId,
endpointId: fixture.endpointId, conversationId: fixture.conversationId, deliveryId: fixture.deliveryId,
kind: "github_review_publication", status: "processed", providerActionId: `github_publication:${fixture.issueId}:comment:${randomUUID()}`,
payload: { operation: "comment", session: { companyId: fixture.companyId, issueId: fixture.issueId,
runId: fixture.runId, agentId: fixture.agentId }, body: "Quantity is ignored by the current total calculation." },
result: { id: "123", url: "https://github.com/paperclip/test-repository/pull/1#issuecomment-123" },
}).returning();
return reply!;
}
it.each([undefined, null] as const)("derives an omitted inherited review wait only after this author reply is delivered: %s", async continuation => {
const fixture = await seedInheritedReviewCompletionTurn(false);
await seedCurrentAuthorReply(fixture);
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
const resolved = await resolveCompletionContinuation({ ...fixture.result, continuation }, fixture.runId,
() => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} })) as PrpStructuredRunResult;
expect(resolved).toMatchObject({ continuation: { kind: "response_wake" } });
await expect(nativeCompletionFeedback(db, fixture.runId, resolved)).resolves.toContain("accepted");
await fixture.port.completeRun({ result: resolved, terminal: fixture.terminal, callerResultId: `wait-result-${fixture.runId}` });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0]).toMatchObject({ status: "in_review" });
expect((await db.select().from(statusDecisions).where(eq(statusDecisions.runId, fixture.runId)))[0])
.toMatchObject({ reasonCode: "external_chat_response_waiting", decisionJson: { effects: [] } });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
});
it.each(["missing_parent", "wrong_cause", "principal_revoked", "new_head", "pending_publication"] as const)(
"does not derive an inherited tree resume wait after %s", async gate => {
const fixture = await seedInheritedReviewCompletionTurn(true);
if (gate === "missing_parent" || gate === "wrong_cause") await db.update(runIdentityContexts)
.set(gate === "missing_parent" ? { parentContextId: null } : { cause: "chat_message" })
.where(eq(runIdentityContexts.runId, fixture.runId));
if (gate === "principal_revoked") await db.update(companyMemberships).set({ status: "suspended" })
.where(eq(companyMemberships.companyId, fixture.companyId));
if (gate === "new_head") await db.insert(chatGitHubReviews).values({ ...fixture.review, id: randomUUID(),
runId: null, deliveryId: randomUUID(), headSha: "c".repeat(40), state: "queued", assessment: null,
createdAt: new Date(Date.now() + 1000) });
if (gate === "pending_publication") await db.update(chatActions).set({ status: "received" })
.where(eq(chatActions.companyId, fixture.companyId));
const context = new PaperclipRunnerToolAuthority(db, fixture)
.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} });
if (gate === "principal_revoked") await expect(context).rejects.toThrow("task is no longer available");
else expect(await context).toMatchObject({ completionContinuation: null });
});
it.each(["missing", "pending", "unknown", "mixed_unknown", "mixed_missing_assessment", "other_run", "other_source", "ack_only", "partial_assessment", "new_head", "revoked"] as const)(
"does not derive an inherited author reply wait after %s", async gate => {
const fixture = await seedInheritedReviewCompletionTurn(false);
const reply = await seedCurrentAuthorReply(fixture);
if (gate === "missing") await db.delete(chatActions).where(eq(chatActions.id, reply.id));
if (gate === "pending" || gate === "unknown") await db.update(chatActions)
.set({ status: gate === "pending" ? "received" : "failed", result: { deliveryUnknown: true } })
.where(eq(chatActions.id, reply.id));
if (gate === "mixed_unknown") await db.insert(chatActions).values({ ...reply, id: randomUUID(),
providerActionId: `github_publication:${randomUUID()}`, status: "failed", result: { deliveryUnknown: true } });
if (gate === "mixed_missing_assessment") await db.insert(chatActions).values({ ...reply, id: randomUUID(),
providerActionId: `github_publication:${randomUUID()}`, payload: { ...reply.payload,
operation: "assessment", reviewId: fixture.review.id }, result: { published: 1, total: 2 } });
if (gate === "other_run") await db.update(chatActions).set({ payload: { ...reply.payload,
session: { companyId: fixture.companyId, issueId: fixture.issueId, runId: fixture.predecessorId, agentId: fixture.agentId } } })
.where(eq(chatActions.id, reply.id));
if (gate === "other_source") {
const [source] = await db.select().from(chatDeliveries).where(eq(chatDeliveries.id, fixture.deliveryId));
const [other] = await db.insert(chatDeliveries).values({ ...source!, id: randomUUID(), providerEventId: randomUUID(), deduplicationKey: randomUUID() }).returning();
await db.update(chatActions).set({ deliveryId: other!.id }).where(eq(chatActions.id, reply.id));
}
if (gate === "partial_assessment") {
const [partial] = await db.insert(chatGitHubReviews).values({ ...fixture.review, id: randomUUID(),
runId: fixture.runId, deliveryId: randomUUID(), createdAt: new Date(Date.now() - 86_400_000), publicationReceipts: {
partial: { id: "123", url: "https://github.com/paperclip/test-repository/pull/1#issuecomment-123", digest: "partial" },
} }).returning();
await db.update(chatActions).set({ status: "failed", payload: { ...reply.payload,
operation: "assessment", reviewId: partial!.id }, result: { code: "publication_failed" } })
.where(eq(chatActions.id, reply.id));
}
if (gate === "ack_only") await db.update(chatActions).set({ kind: "github_response_comment" })
.where(eq(chatActions.id, reply.id));
if (gate === "new_head") await db.insert(chatGitHubReviews).values({ ...fixture.review, id: randomUUID(),
runId: null, deliveryId: randomUUID(), headSha: "c".repeat(40), state: "queued", assessment: null,
createdAt: new Date(Date.now() + 1000) });
if (gate === "revoked") await db.update(toolConnections).set({ enabled: false })
.where(eq(toolConnections.companyId, fixture.companyId));
const context = await new PaperclipRunnerToolAuthority(db, fixture)
.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} });
expect(context).toMatchObject({ completionContinuation: null });
});
it.each(["current", "resume", "reply"] as const)("projects only bound authority while deriving a published GitHub completion wait: %s", async turn => {
const fixture = turn === "current" ? await seedPublishedReviewCompletionTurn() : await seedInheritedReviewCompletionTurn(turn === "resume");
const reply = turn === "reply" ? await seedCurrentAuthorReply(fixture as Awaited<ReturnType<typeof seedInheritedReviewCompletionTurn>>) : null;
const [run] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
const payload = "unused-saved-review-rendering".repeat(32768);
await db.update(heartbeatRuns).set({ runnerProfileJson: { retained: payload }, stdoutExcerpt: payload, stderrExcerpt: payload,
contextSnapshot: { ...run.contextSnapshot, executionContinuation: { savedInput: payload },
paperclipTaskMarkdown: payload, paperclipWake: { ...(run.contextSnapshot!.paperclipWake as object), retainedPayload: payload } },
}).where(eq(heartbeatRuns.id, fixture.runId));
if (reply) await db.update(chatActions).set({ result: { ...reply.result, savedProviderOutput: payload } }).where(eq(chatActions.id, reply.id));
const observed = createDb(temporary?.connectionString ?? externalTestDatabaseUrl!);
const events: Array<Record<string, unknown>> = [], now = Date.now();
const capture = createDatabaseResponseCapture({ start: now - 100, end: now + 59_000 }, event => events.push(event));
installDatabaseResponseCapture(observed, { heartbeatRuns, chatActions }, capture);
try {
const context = await new PaperclipRunnerToolAuthority(observed, fixture)
.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} });
expect(context).toMatchObject({ completionContinuation: { kind: "response_wake" } });
capture.flush();
const reads = events.filter(event => event.table === "heartbeat_runs" && event.operation === "select");
expect(reads.length).toBeGreaterThan(0);
expect(reads.reduce((bytes, event) => bytes + Number(event.responseValueBytes), 0)).toBeLessThan(16 * 1024);
const actionReads = events.filter(event => event.table === "chat_actions" && event.operation === "select");
expect(actionReads.reduce((bytes, event) => bytes + Number(event.responseValueBytes), 0)).toBeLessThan(16 * 1024);
expect(JSON.stringify(context)).not.toContain(payload);
} finally { capture.flush(); await observed.$client.end(); }
});
it.each([undefined, null, "malformed-first"] as const)("derives an omitted published GitHub review wait through current task context: %s", async absent => {
const fixture = await seedPublishedReviewCompletionTurn();
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const read = () => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} });
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
const report = { ...fixture.result, continuation: absent === "malformed-first" ? "response_wake" : absent };
if (absent === "malformed-first") {
// Explicit malformed input remains invalid; the next omitted report can
// recover from authoritative state without changing its disposition.
expect(await resolveCompletionContinuation(report, fixture.runId, read)).toEqual(report);
expect(validatePrpStructuredRunResult(report).ok).toBe(false);
delete report.continuation;
}
const resolved = await resolveCompletionContinuation(report, fixture.runId, read) as PrpStructuredRunResult;
expect(resolved).toMatchObject({ reportedWorkDisposition: "yielded", continuation: { kind: "response_wake" } });
await expect(nativeCompletionFeedback(db, fixture.runId, resolved)).resolves.toContain("accepted");
await fixture.port.completeRun({ result: resolved, terminal: fixture.terminal, callerResultId: `wait-result-${fixture.runId}` });
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0]).toMatchObject({ status: "in_review" });
expect((await db.select().from(statusDecisions).where(eq(statusDecisions.runId, fixture.runId)))[0])
.toMatchObject({ reasonCode: "external_chat_response_waiting", decisionJson: { effects: [] } });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
});
it.each(["missing_publication", "pending_publication", "unknown_publication", "new_head", "other_run", "review_mismatch", "principal_revoked", "connection_revoked", "endpoint_revoked", "ownership_changed"] as const)(
"does not derive a published GitHub completion wait after %s", async gate => {
const fixture = await seedPublishedReviewCompletionTurn();
if (gate === "missing_publication") await db.delete(chatActions).where(eq(chatActions.companyId, fixture.companyId));
if (gate === "pending_publication" || gate === "unknown_publication") await db.update(chatActions)
.set({ status: gate === "pending_publication" ? "received" : "failed", result: { deliveryUnknown: true } })
.where(eq(chatActions.companyId, fixture.companyId));
if (gate === "new_head") await db.insert(chatGitHubReviews).values({ ...fixture.review, id: randomUUID(),
runId: null, deliveryId: randomUUID(), headSha: "c".repeat(40), state: "queued", assessment: null,
createdAt: new Date(Date.now() + 1000) });
if (gate === "other_run") {
const [run] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, fixture.runId));
const previousId = randomUUID();
await db.insert(heartbeatRuns).values({ ...run!, id: previousId, activeIdentityContextId: null,
status: "succeeded", finishedAt: new Date() });
await db.update(chatGitHubReviews).set({ runId: previousId }).where(eq(chatGitHubReviews.id, fixture.review.id));
}
if (gate === "review_mismatch") await db.update(chatGitHubReviews).set({ assessment: {
...fixture.review.assessment!, reviewedCommit: "f".repeat(40),
} }).where(eq(chatGitHubReviews.id, fixture.review.id));
if (gate === "principal_revoked") await db.update(companyMemberships).set({ status: "suspended" })
.where(eq(companyMemberships.companyId, fixture.companyId));
if (gate === "connection_revoked") await db.update(toolConnections).set({ enabled: false })
.where(eq(toolConnections.companyId, fixture.companyId));
if (gate === "endpoint_revoked") await db.update(chatEndpoints).set({ status: "revoked" })
.where(eq(chatEndpoints.id, fixture.endpointId));
if (gate === "ownership_changed") await db.update(issues).set({ executionRunId: null })
.where(eq(issues.id, fixture.issueId));
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
await expect(resolveCompletionContinuation({ ...fixture.result, continuation: undefined }, fixture.runId,
() => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} }))).rejects.toThrow();
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
},
);
it("serializes an actual connection disable against the final status committer proof", async () => {
const fixture = await seedPublishedReviewCompletionTurn();
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
const result = await resolveCompletionContinuation({ ...fixture.result, continuation: null }, fixture.runId,
() => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} })) as PrpStructuredRunResult;
await nativeCompletionFeedback(db, fixture.runId, result);
await fixture.port.completeRun({ result, terminal: fixture.terminal, callerResultId: `wait-result-${fixture.runId}` });
const [connection] = await db.select().from(toolConnections).where(eq(toolConnections.companyId, fixture.companyId));
const applicationName = `review-wait-race-${fixture.runId}`;
const writerDb = createDb(externalTestDatabaseUrl ?? temporary!.connectionString, { applicationName });
let writer: Promise<unknown> | undefined;
let writerFinished = false, serialized = false;
const readProof = publishedReviewAuthority.readInheritedGitHubReviewWait;
const readForCommit = chatWaitAuthority.readPublishedGitHubReviewCompletionContinuationInTransaction;
const commitGuard = vi.spyOn(chatWaitAuthority, "readPublishedGitHubReviewCompletionContinuationInTransaction")
.mockImplementationOnce(async (...args) => {
const [committer] = await args[0].execute(sql`select pg_backend_pid() as pid`);
const proofGuard = vi.spyOn(publishedReviewAuthority, "readInheritedGitHubReviewWait")
.mockImplementationOnce(async (...proofArgs) => {
const proof = await readProof(...proofArgs);
// This is the same maintained writer used by PATCH tool-connections.
// It does not borrow the task's issue lock to disable a chat connection.
writer = toolAccessService(writerDb).updateConnection(connection!.id, { enabled: false })
.finally(() => { writerFinished = true; });
await vi.waitFor(async () => {
const waiting = await db.execute(sql`select query from pg_stat_activity
where application_name = ${applicationName} and wait_event_type = 'Lock'`);
expect(writerFinished || waiting.length > 0).toBe(true);
if (waiting.length) expect(String(waiting[0]!.query)).toContain('update "tool_connections"');
serialized = !writerFinished && waiting.length > 0;
if (serialized) {
const blockers = await db.execute(sql`select pg_blocking_pids(pid) as blockers from pg_stat_activity where application_name = ${applicationName} and wait_event_type = 'Lock'`);
expect(blockers[0]!.blockers).toContain(Number(committer!.pid));
}
});
return proof;
});
try { return await readForCommit(...args); }
finally { proofGuard.mockRestore(); }
});
try {
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
await writer;
expect(serialized).toBe(true);
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0]).toMatchObject({ status: "in_review" });
expect((await db.select().from(toolConnections).where(eq(toolConnections.id, connection!.id)))[0]).toMatchObject({ enabled: false });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
} finally { commitGuard.mockRestore(); await Promise.allSettled(writer ? [writer] : []); await writerDb.$client.end({ timeout: 1 }); }
});
it.each((["current", "resume", "reply"] as const).flatMap(turn =>
(["before_feedback", "before_finalization", "during_commit"] as const).map(phase => ({ turn, phase }))))(
"rechecks a projected GitHub completion wait $turn $phase", async ({ turn, phase }) => {
const fixture = turn === "current" ? await seedPublishedReviewCompletionTurn() : await seedInheritedReviewCompletionTurn(turn === "resume");
const reply = turn === "reply" ? await seedCurrentAuthorReply(fixture as Awaited<ReturnType<typeof seedInheritedReviewCompletionTurn>>) : null;
const authority = new PaperclipRunnerToolAuthority(db, fixture);
const { resolveCompletionContinuation } = await import(new URL("../../../../packages/paperclip-runner/src/drivers/completion-continuation.ts", import.meta.url).href);
const result = await resolveCompletionContinuation({ ...fixture.result, continuation: null }, fixture.runId,
() => authority.execute({ tool: "get_task_context", callId: randomUUID(), arguments: {} })) as PrpStructuredRunResult;
const revoke = () => reply
? db.update(chatActions).set({ status: "failed", result: { deliveryUnknown: true } }).where(eq(chatActions.id, reply.id))
: turn === "resume" ? db.update(chatActions).set({ status: "received" }).where(eq(chatActions.companyId, fixture.companyId))
: db.update(toolConnections).set({ enabled: false }).where(eq(toolConnections.companyId, fixture.companyId));
if (phase === "before_feedback") {
await revoke();
await expect(nativeCompletionFeedback(db, fixture.runId, result)).rejects.toThrow("no longer authorized");
return;
}
await nativeCompletionFeedback(db, fixture.runId, result);
await fixture.port.completeRun({ result, terminal: fixture.terminal, callerResultId: `wait-result-${fixture.runId}` });
let guard: ReturnType<typeof vi.spyOn> | undefined;
if (phase === "before_finalization") await revoke();
if (phase === "during_commit") {
const read = chatWaitAuthority.readPublishedGitHubReviewCompletionContinuationInTransaction;
guard = vi.spyOn(chatWaitAuthority, "readPublishedGitHubReviewCompletionContinuationInTransaction")
.mockImplementation(async (...args) => {
if (reply) await args[0].update(chatActions).set({ status: "failed", result: { deliveryUnknown: true } }).where(eq(chatActions.id, reply.id));
else if (turn === "resume") await args[0].update(chatActions).set({ status: "received" }).where(eq(chatActions.companyId, fixture.companyId));
else await args[0].update(toolConnections).set({ enabled: false }).where(eq(toolConnections.companyId, fixture.companyId));
return read(...args);
});
}
try {
const finalized = await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect((await db.select().from(issues).where(eq(issues.id, fixture.issueId)))[0]).toMatchObject({ status: "in_progress" });
const decisions = await db.select().from(statusDecisions).where(eq(statusDecisions.runId, fixture.runId));
if (phase === "during_commit") {
expect(finalized).toMatchObject({ phase: "retryable_failure" });
expect(decisions).toHaveLength(0);
} else expect(decisions.at(-1)).toMatchObject({ reasonCode: "external_chat_response_wait_authorization_lost" });
expect(await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.companyId, fixture.companyId))).toHaveLength(0);
} finally { guard?.mockRestore(); }
},
);
it("admits a real repository-access request to Mine until it is resolved", async () => {
const fixture = await seedWaitTurn("github", [{ kind: "approval", ownerClass: "human", summary: "Grant repository access so I can read the diff." }], {
reportedWorkDisposition: "needs_review", continuation: undefined,
});
await db.update(issues).set({ responsibleUserId: fixture.userId }).where(eq(issues.id, fixture.issueId));
await finalizeNativeRun({ db, runId: fixture.runId, workspaceFinalizeStatus: "succeeded", projectRunStatus: true });
expect((await issueService(db).list(fixture.companyId, { touchedByUserId: fixture.userId })).map(issue => issue.id)).toContain(fixture.issueId);
const [request] = await db.select().from(issueThreadInteractions).where(eq(issueThreadInteractions.issueId, fixture.issueId));
expect(request).toMatchObject({ status: "pending", addresseeUserId: fixture.userId });
await issueThreadInteractionService(db).withdrawInteraction(
{ id: fixture.issueId, companyId: fixture.companyId }, request!.id,
{ reason: "Repository access restored." }, { userId: fixture.userId },
);
expect(await issueService(db).list(fixture.companyId, { touchedByUserId: fixture.userId })).toEqual([]);
});
it("does not schedule a fallback run when current chat permission is revoked", async () => {
const fixture = await seedWaitTurn();
await db
.update(companyMemberships)
.set({ status: "suspended", updatedAt: new Date() })
.where(eq(companyMemberships.principalId, fixture.userId));
await finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
await expect(
db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({
reasonCode: "external_chat_response_wait_authorization_lost",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
await expect(
db
.select()
.from(agentWakeupRequests)
.where(
and(
eq(agentWakeupRequests.companyId, fixture.companyId),
eq(agentWakeupRequests.agentId, fixture.agentId),
),
),
).resolves.toEqual([]);
});
it("retries a contended endpoint proof without deadlocking or scheduling work", async () => {
const fixture = await seedWaitTurn();
let releaseEndpoint!: () => void;
let endpointLocked!: () => void;
const endpointRelease = new Promise<void>((resolve) => {
releaseEndpoint = resolve;
});
const endpointLockObserved = new Promise<void>((resolve) => {
endpointLocked = resolve;
});
const endpointHolder = db.transaction(async (tx) => {
await tx
.select({ id: chatEndpoints.id })
.from(chatEndpoints)
.where(eq(chatEndpoints.id, fixture.endpointId))
.for("update");
endpointLocked();
await endpointRelease;
});
await endpointLockObserved;
const finalization = finalizeNativeRun({
db,
runId: fixture.runId,
workspaceFinalizeStatus: "succeeded",
projectRunStatus: true,
});
let contentionAssertionError: unknown = null;
try {
let firstArbitrationUpdatedAt = 0;
await vi.waitFor(
async () => {
const [coordinator] = await db
.select({
assessmentId: nativeRunFinalizations.assessmentId,
phase: nativeRunFinalizations.phase,
updatedAt: nativeRunFinalizations.updatedAt,
})
.from(nativeRunFinalizations)
.where(eq(nativeRunFinalizations.runId, fixture.runId));
expect(coordinator).toEqual(
expect.objectContaining({
assessmentId: expect.any(String),
phase: "arbitrating",
}),
);
firstArbitrationUpdatedAt = coordinator!.updatedAt.getTime();
},
{ timeout: 1_000, interval: 5 },
);
await vi.waitFor(
async () => {
const [coordinator] = await db
.select({ updatedAt: nativeRunFinalizations.updatedAt })
.from(nativeRunFinalizations)
.where(eq(nativeRunFinalizations.runId, fixture.runId));
expect(coordinator!.updatedAt.getTime()).toBeGreaterThan(
firstArbitrationUpdatedAt,
);
},
{ timeout: 1_000, interval: 5 },
);
} catch (error) {
contentionAssertionError = error;
} finally {
releaseEndpoint();
}
await endpointHolder;
if (contentionAssertionError) throw contentionAssertionError;
await finalization;
await expect(
db
.select()
.from(statusDecisions)
.where(eq(statusDecisions.issueId, fixture.issueId)),
).resolves.toEqual([
expect.objectContaining({
reasonCode: "external_chat_response_waiting",
decisionJson: expect.objectContaining({ effects: [] }),
}),
]);
await expect(
db
.select()
.from(agentWakeupRequests)
.where(
and(
eq(agentWakeupRequests.companyId, fixture.companyId),
eq(agentWakeupRequests.agentId, fixture.agentId),
),
),
).resolves.toEqual([]);
});
});