Files
PaperClipAI/server/src/services/native-runtime/native-execution-input.ts
T
DottaandPaperclip 992f720262 fix: make runner task context ownership explicit (#13753)
<!-- Write all pull request text in Simplified Technical English
(ASD-STE100). -->

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - Task descriptions, comments, continuation data, skills, and
execution rules enter several agent adapters.
> - The same source can be rendered by more than one automatic input
carrier.
> - Failed resumes can also rebuild input from stale or compact context.
> - This pull request gives each Paperclip-owned source one delivery
owner and preserves the required transport boundaries.
> - It adds deterministic adapter, interaction, runner, and browser
tests for these boundaries.
> - The benefit is more predictable context delivery with explicit
evidence for later live qualification.

## Linked Issues or Issue Description

Related: #13144 removes a duplicate environment payload and bounds wake
lists. Related: #11360 addresses Hermes resume behavior. This pull
request preserves compatible active-session formats while repairing
context ownership and stale question creation.

**What happened?**

Task descriptions and comments could enter more than one automatic
context block. Native transports could wrap a complete model input in a
second task envelope. Some legacy and gateway adapters could omit the
owned assignment on ordinary tasks or rebuild a failed resume with stale
compact context. A continuation could also request a question after
newer human comments had arrived.

**Expected behavior**

Each task or comment source has one automatic model-facing owner.
Distinct comment IDs and repeated wording remain distinct. Fresh
fallback attempts rebuild the required full context. A question request
is rejected when newer queued human direction makes it stale. Harness
access policy remains owned by execution configuration.

**Steps to reproduce**

1. Build a task with a description and current comments.
2. Capture the actual adapter or runner input.
3. Compare source ownership and task-envelope nesting.
4. Queue a human comment before a continuation requests a question.
5. Trigger a failed resume and inspect the fresh retry input.
6. Run the focused adapter, interaction, runner, and browser checks.

## What Changed

- Add shared prompt-section selection at the provider-attempt boundary.
- Deliver owned assignment context through native, legacy CLI, ACP,
gateway, cloud, Pi, Kimi, Grok, Gemini, OpenCode, Cursor, OpenClaw, and
Hermes paths.
- Rebuild full or compact context after resume recovery changes the
attempt. Add native and Claude ACP tests of actual recovery requests.
- Preserve custom templates, loaded instruction files, execution
policies, and older active-session formats.
- Record continuation source metadata and reject stale question creation
under the issue-row lock.
- Add explicit Product E2E context-integrity profiles, prerequisite
gates, credential-isolation checks, and report fixtures.
- Bypass service-worker forwarding for same-origin Vite development
modules. A real Chromium test fails with resource exhaustion before the
repair and passes after it. Production asset caching keeps its existing
policy.
- Add browser diagnostics and service-worker module-loading regressions.
- Add an explicit zero-retry eval option. The default retry behavior
remains unchanged. Each campaign records its effective policy.
- Remove the model-facing working-directory sentence from four prompt
builders. Existing workspace, sandbox, permission, and custom-template
configuration remains unchanged.
- Align the everyday workflow assertion with the current 47-entry
catalog.

Compared with current upstream master, the branch carries the
context-ownership implementation and its tests, the explicit
context-integrity catalog and evidence harness, and the focused browser
regression checks.

## Verification

**Merge assessment:** focused regression evidence supports merge. This
is not full completion of the original broad qualification matrix. The
maintainer has authorized merge after fresh verification of the master
integration.

- Current head: `bbd52f82114eabf09bc7b1a7e97d54a5b43bbc00`. This
integrates current master `2f585ef26a1814fa209715242d1ca791b63e4c4e`.
All 14 conflicts are resolved. Cancellation checks, workspace
finalization, native Grok support, and both sets of tests are retained.
- Current-head Greptile: **5/5**, with no blocking findings. The review
names this exact commit. All **59 reported checks are terminal: 55
successful, 4 skipped, zero pending or failing**. This includes the full
root general and serialized suites, separate runner checks, typecheck,
build, canary, browser E2E, Docker, and security checks. The successful
legacy security status is included in that total.
- After integration: workspace typecheck and full build passed. Separate
runner checks passed: **2,160 TypeScript tests (10 skipped), 582 Rust
tests, and 39 preparation checks**. Other passing checks include 621
Product E2E harness units, 376 focused shared/adapter tests, 160
real-database/API tests, 86 Hermes tests, 18 browser-support checks, and
Product E2E typechecking. The complete root suite passed in CI. The
duplicate local monolithic root run was stopped after that CI result; it
is not counted as a completed local pass.
- New native recovery coverage retains full assignment, completion
contract, and explicit skill selection after safe replacement, for old
and prepared input formats. Full native session test file: **136/136
passed**.
- New Claude ACP coverage captures actual fresh, resumed, and
missing-session fallback requests. It verifies one assignment copy,
comment order, identical text under distinct comment IDs, and full
fallback context. Full file: **33/33 passed**. Both affected TypeScript
checks passed.
- Existing deterministic tests cover source revisions, approval and
trust boundaries, completion validation, custom templates, compatible
sessions, standalone driver wrapping, and maintained adapter transport
requests.
- Provider-free browser support: **17/17 passed** after the master
merge. Service-worker unit tests: **33/33 passed**. The module-overload
regression failed before the repair and passed after it in real
Chromium.

### Fresh live comparisons

The new batch ran exactly four Product E2E attempts. **All four passed
on the first attempt; no retries.** Each has six terminal matchers plus
the existing browser lifecycle and invariant checks.

| Exact case ID | Control | Candidate |
|---|---|---|
| `core-compatibility.runner-codex.local.plan-revise-accept` | Passed |
Passed |
|
`local-session-integrity.runner-acpx-claude.local.structured-question-restart-resume`
| Passed | Passed |

The plan case checks a revised canonical plan and revision-bound
approval before completion. The question case restarts the server before
submitting the answer, then verifies the continuation completes.

Control source is `dfa4e1bda8d50a1a01746603251a9128dbe9d0d6`. Candidate
source is `79fcdb5dece501d28064ea9da306603881b46f0c`. They use identical
frozen definitions and provider versions: Codex `0.156.0` with
`gpt-5.6-sol`; ACPX `0.13.1` / Claude ACP `0.73.0` with
`claude-sonnet-5`. The September 24 head added master browser recovery
and test-only changes. The September 28 head also integrates newer
master changes, including cancellation, workspace finalization, and
native Grok. These are frozen-source live results, not exact-head live
runs.

The candidate received one description copy where the control initially
received three. The submitted initial plan envelopes were 7,969 versus
19,097 characters. Question envelopes were 7,592 versus 18,919. These
are structural measurements, not whole-provider token or dollar savings.

### Earlier evidence and failed attempts

- The preceding fresh batch has four effective passing pairs: OpenCode
comment continuation and assigned skill, native Codex comment
continuation, and native Claude comment continuation. It retains **11
attempts: eight passed and three failed**.
- Original failures remain recorded: missing local PostgreSQL library
links before task creation; host-sleep cleanup after task/page checks
passed; and a Claude **control** session-open rejection before a model
turn. Setup was repaired identically on both worktrees. The permitted
unchanged infrastructure retries passed. The underlying Claude provider
startup error was not retained and remains unknown.
- Older R2 retains **17 passes and one failure** across 18 attempts,
including eight both-pass native/legacy Codex/Claude pairs. Its OpenCode
blank-page failure led to the service-worker repair. R2 is historical
evidence: master changed the native fixed prompt and removed duplicate
wake environment data afterward.
- The September 24 CI run initially failed one unrelated preview
readiness test (`ECONNREFUSED` on its local fixture). Its test and
production code match master. Isolated local verification passed **28
tests, 3 skipped**. One unchanged CI retry passed the full shard: **831
passed, 1 skipped**, including all **31 preview-exposure tests**. The
aggregate CI gate passed afterward. The precise startup cause remains
unknown; a port race is a hypothesis, not a proved cause.

### Limits

The original wider profile/workflow matrix, repeated trials, and remote
Daytona qualification are incomplete. These results support a focused
merge recommendation, not statistical equivalence or universal harness
qualification. Some usage receipts are missing in both variants, so no
token or dollar savings are claimed. The $500 ceiling was preserved
using conservative allowances; failed attempts and unknown charges
remain in the ledger.

Reproduce the focused additions with `pnpm exec vitest run
packages/adapters/claude-local/src/server/acp.test.ts` and `pnpm
--filter @paperclipai/paperclip-runner exec vitest run
src/native-session-runtime.test.ts`. Full checks use `pnpm -r
typecheck`, `pnpm test:run`, `pnpm build`, and the separate runner
checks. Paid evals require the frozen definitions, profiles, and
credentials; do not use `--all` as a substitute for the selected cases.

## Risks

- Context placement changes can affect model behavior. Deterministic
checks cover the selected paths, but live qualification remains
incomplete.
- The stale-question guard can reject a request when queued human
comments arrived during the run. This is intended.
- New stored inputs and model envelopes retain compatibility readers for
older active sessions.
- Custom templates may intentionally repeat content.
- Removing a model-facing working-directory sentence does not change
filesystem, command, sandbox, or permission configuration.
- The worker bypass applies only to same-origin development module
paths. Cache-policy tests preserve private-response handling and
production asset caching. Mounted HTTP fixture changes remain test-only.
- This PR does not claim measured token savings or statistical
equivalence across every harness.

## Model Used

OpenAI Codex, exact model gpt-6-astra, with repository tools and code
execution. Bounded supporting work used gpt-5.6-luna and gpt-6-luna. The
serving context-window size is not exposed in this task.

## 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 described the issue in-PR using the required issue fields
- [x] I have not referenced internal/instance-local Paperclip issues or
links
- [x] My branch name describes the change and contains no internal
ticket id
- [x] I have run the focused local checks and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect these changes
- [x] I have considered and documented risks above
- [x] All current-head Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
for the current head
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-09-28 14:49:14 -05:00

318 lines
13 KiB
TypeScript

import type { PaperclipTurnContext } from "@paperclipai/adapter-utils/server-utils";
import { createHash } from "node:crypto";
import { buildNativeContinuationPrompt } from "./native-continuation.js";
import type {
NativeAcpxAgent,
NativeAcpxPermissionMode,
NativeCodexApprovalPolicy,
NativeExecutionInputV5,
NativeCompletionSource,
NativeInteractionResponseEnvelope,
NativeOpenCodePermissionMode,
NativePlanningContext,
NativeRuntimeContextSnapshot,
StrictCompletionContractInput,
} from "../../vendor/paperclip-runner/index.js";
import {
parseNativeExecutionInput,
resolveQualifiedAcpxProfile,
} from "../../vendor/paperclip-runner/index.js";
import {
isPaperclipExternalChatContractTurn,
isPaperclipExternalChatQuestionResponseTurn,
selectPaperclipPromptSections,
} from "@paperclipai/adapter-utils/server-utils";
const NATIVE_GITHUB_ATTACHMENT_RECOVERY_GUIDANCE = [
"## GitHub attachment recovery navigation",
"Paperclip owns recovery navigation for unavailable GitHub attachments. It may append an authenticated task link after an accepted response, only when the current source remains authorized and a safe configured Board URL is available. The model does not select or authorize that link.",
"A task URL missing from your prompt or tool results is not evidence that no task link can be provided; do not claim that a link is unavailable merely because you cannot see its URL. Do not invent a URL or promise that a link will appear. Briefly explain the unavailable input and ask the user to attach it directly to this Paperclip task or paste the needed text. Never infer the file's contents or substitute an older file.",
].join("\n");
/** Closed constructor: callers cannot spread legacy context or environment data. */
export function buildNativeExecutionInput(input: {
companyId: string;
runId: string;
issue: {
id: string;
identifier: string | null;
title: string;
description: string | null;
workMode: string;
};
taskPrompt: string;
initialCommunicationGuidance?: string | null;
/**
* The already-sanitized Paperclip wake envelope for this run. Native drivers
* receive a closed execution input rather than the legacy adapter context,
* so the constructor must deliberately project the same bounded wake delta
* that legacy adapters place in their provider prompt.
*/
wakePayload?: unknown;
/** Additive source ownership emitted by the server task/wake builders. */
turnContext?: unknown;
resumedSession?: boolean;
previousTurn?: { runId: string; task: { title: string; description: string | null } } | null;
conversationMode?: boolean;
agentId: string;
workspace: {
id: string;
cwd: string;
repoUrl: string | null;
repoRef: string | null;
branchName: string | null;
};
normalizedSessionId: string | null;
provider?: "codex" | "opencode" | "claude_managed" | "aws_agentcore" | "acpx";
acpxAgent?: NativeAcpxAgent;
codexApprovalPolicy?: NativeCodexApprovalPolicy;
opencodePermissionMode?: NativeOpenCodePermissionMode;
acpxPermissionMode?: NativeAcpxPermissionMode;
model?: string | null;
managedProfile?: Extract<
NativeExecutionInputV5["provider"],
{ kind: "claude_managed" }
>["managedProfile"];
maxSessionListCostUsd?: number;
agentCoreProfile?: Extract<
NativeExecutionInputV5["provider"],
{ kind: "aws_agentcore" }
>["agentCoreProfile"];
maxEstimatedSessionCostUsd?: number;
invocationLimits?: Extract<
NativeExecutionInputV5["provider"],
{ kind: "aws_agentcore" }
>["invocationLimits"];
lifecyclePolicy?: NativeExecutionInputV5["session"]["lifecyclePolicy"];
executionMode?: "default" | "plan";
planningContext?: NativePlanningContext | null;
interactionResponses?: NativeInteractionResponseEnvelope[];
completionContract: {
id: string;
sha256: string;
schemaVersion: string;
contract: StrictCompletionContractInput;
sources?: Array<{ id: string; source: NativeCompletionSource }>;
};
runtimeContext: NativeRuntimeContextSnapshot;
}): NativeExecutionInputV5 {
if (input.issue.workMode !== "standard" && input.issue.workMode !== "planning" && input.issue.workMode !== "ask") {
throw new Error("native_execution_input_invalid: issue work mode must be standard, planning, or ask");
}
const executionMode = input.executionMode
?? (input.issue.workMode === "planning" ? "plan" : "default");
const acpxProfile = input.provider === "acpx"
? resolveQualifiedAcpxProfile(
input.acpxAgent ?? "codex",
input.model ?? "",
)
: null;
// Answers are materialized from the authoritative interaction only for this
// invocation; do not persist a duplicate answer in the durable wake snapshot.
const wake =
input.wakePayload &&
typeof input.wakePayload === "object" &&
!Array.isArray(input.wakePayload)
? (input.wakePayload as Record<string, unknown>)
: null;
const question = wake?.externalChatQuestionResponse
? input.interactionResponses?.find(
(response) =>
response.interactionId === wake.interactionId &&
response.kind === "ask_user_questions" &&
response.response.status === "answered",
)
: null;
const answerResult = question?.response.result as
Record<string, unknown> | undefined;
// The server supplies only the revalidated answer chain, in source order.
// Keep prior choices available even when this continuation starts a fresh
// provider session; never recover them from model prose or a transcript.
const answerChain = question
? input.interactionResponses?.filter((response) =>
response.kind === "ask_user_questions" &&
response.response.status === "answered" &&
typeof (response.response.result as Record<string, unknown> | undefined)
?.summaryMarkdown === "string")
: null;
const answerSummary =
answerChain && answerChain.length > 1 &&
answerChain.at(-1)?.interactionId === question?.interactionId
? answerChain.map((response, index) => {
const label = index === answerChain.length - 1
? "Latest answered question" : `Earlier answer ${index + 1}`;
return `${label}:\n${(response.response.result as Record<string, unknown>).summaryMarkdown}`;
}).join("\n\n")
: answerResult?.summaryMarkdown;
const wakePayload =
question && typeof answerSummary === "string"
? {
...wake,
questionResponse: {
interactionId: question.interactionId,
summaryMarkdown: answerSummary,
},
}
: input.wakePayload;
// Build the full bootstrap through the same owner as legacy adapters.
// Verified native resume selection stays at the existing session boundary.
const { taskContextNote, wakePrompt } = selectPaperclipPromptSections({
paperclipTaskMarkdownAssignment: input.taskPrompt,
paperclipWake: wakePayload,
conversationMode: input.conversationMode,
}, {
resumedSession: false,
includeCommunicationGuidance: false,
nativeWakeReaderAvailable: true,
});
const externalChatTurn =
isPaperclipExternalChatContractTurn(wakePayload) ||
isPaperclipExternalChatQuestionResponseTurn(wakePayload);
const taskPrompt = [
wakePrompt,
// Durable task questions must survive the current provider turn.
// Keep routing visible before deferred tool discovery; usage belongs in the tool schema.
"Use Paperclip's request_human_input for durable task questions.",
externalChatTurn && wake?.externalChatProvider === "github"
? NATIVE_GITHUB_ATTACHMENT_RECOVERY_GUIDANCE
: "",
taskContextNote,
]
.filter((section) => section.length > 0)
.join("\n\n");
const completionSources = !externalChatTurn && !input.conversationMode && input.taskPrompt.trim()
? verifiedCompletionSources(input.turnContext, input.completionContract.sources ?? [])
: [];
return parseNativeExecutionInput({
schema: "paperclip.native-execution-input.v5",
...(input.initialCommunicationGuidance ? { initialCommunicationGuidance: input.initialCommunicationGuidance } : {}),
...(input.resumedSession && input.previousTurn && !input.conversationMode ? {
continuationPrompt: buildNativeContinuationPrompt({
wakePayload: input.wakePayload,
previousRunId: input.previousTurn.runId,
previousIssue: input.previousTurn.task,
allowExternalChat: true,
issue: input.issue,
}),
} : {}),
executionMode,
planningContext: input.planningContext ?? null,
binding: {
companyId: input.companyId,
runId: input.runId,
issueId: input.issue.id,
agentId: input.agentId,
executionWorkspaceId: input.workspace.id,
},
task: {
identifier: input.issue.identifier ?? input.issue.id,
// The issue title is durable background context and may itself contain an
// exact-output instruction from the thread's first message. Repeating it
// as the native turn title can override a newer provider message in small
// models. Keep the canonical title and description in task.prompt as
// explicitly labeled background, but give authenticated external-chat
// turns neutral structured fields.
title: externalChatTurn ? "External chat follow-up" : input.issue.title,
description: externalChatTurn ? null : input.issue.description,
prompt: taskPrompt,
workMode: input.issue.workMode,
},
workspace: {
cwd: input.workspace.cwd,
repoUrl: input.workspace.repoUrl,
repoRef: input.workspace.repoRef,
branchName: input.workspace.branchName,
},
session: {
normalizedSessionId: input.normalizedSessionId,
driverKind: input.provider === "opencode"
? "opencode_server"
: input.provider === "claude_managed"
? "claude_managed_agents_api"
: input.provider === "aws_agentcore"
? "aws_agentcore_harness_api"
: input.provider === "acpx"
? "acpx_runtime"
: "codex_app_server",
protocolVersion: 1,
lifecyclePolicy: input.lifecyclePolicy ?? { mode: "per_turn", idleTimeoutMs: null },
},
provider: input.provider === "claude_managed"
? {
kind: "claude_managed",
model: input.model,
managedProfile: input.managedProfile,
maxSessionListCostUsd: input.maxSessionListCostUsd,
}
: input.provider === "aws_agentcore"
? {
kind: "aws_agentcore",
model: input.model,
agentCoreProfile: input.agentCoreProfile,
maxEstimatedSessionCostUsd: input.maxEstimatedSessionCostUsd,
invocationLimits: input.invocationLimits,
}
: input.provider === "acpx"
? {
kind: "acpx",
agent: acpxProfile!.agent,
model: input.model,
permissionMode: input.acpxPermissionMode ?? "approve-all",
profile: {
driverKind: acpxProfile!.driverKind,
protocolVersion: acpxProfile!.protocolVersion,
acpxVersion: acpxProfile!.acpxVersion,
agent: acpxProfile!.agent,
agentProfileVersion: acpxProfile!.agentProfileVersion,
agentServerPackage: acpxProfile!.agentServerPackage,
agentServerVersion: acpxProfile!.agentServerVersion,
agentRuntimePackage: acpxProfile!.agentRuntimePackage,
agentRuntimeVersion: acpxProfile!.agentRuntimeVersion,
commandDigest: acpxProfile!.commandDigest,
},
}
: input.provider === "opencode"
? {
kind: "opencode",
model: input.model,
permissionMode: input.opencodePermissionMode ?? "allow",
}
: {
kind: "codex",
model: input.model ?? null,
approvalPolicy: input.codexApprovalPolicy ?? "never",
},
completionContract: {
id: input.completionContract.id,
sha256: input.completionContract.sha256,
schemaVersion: input.completionContract.schemaVersion,
contract: input.completionContract.contract,
},
...(completionSources.length ? { completionSources: {
promptSha256: createHash("sha256").update(taskPrompt).digest("hex"),
contractRevision: input.completionContract.contract.revision,
criteria: completionSources,
} } : {}),
interactionResponses: input.interactionResponses ?? [],
credentialBindings: [],
runtimeContext: input.runtimeContext,
}) as NativeExecutionInputV5;
}
/** Verify source identities and revisions, never guess provenance from requirement text. */
function verifiedCompletionSources(
value: unknown,
sources: Array<{ id: string; source: NativeCompletionSource }>,
): Array<{ id: string; source: NativeCompletionSource }> {
if (!value || typeof value !== "object" || Array.isArray(value)) return [];
const context = value as Partial<PaperclipTurnContext>;
if (context.version !== 1) return [];
return sources.filter(({ source }) => {
if (source.kind === "description") {
return context.assignment?.owner === "task_markdown" && context.assignment.description?.id === source.id && context.assignment.description.revision === source.revision;
}
return context.events?.owner === "wake_prompt" && Array.isArray(context.events.comments) && context.events.comments.some((comment) => comment.id === source.id && comment.revision === source.revision);
});
}