Files
PaperClipAI/packages/paperclip-runner/devtools/issue-thread/src/App.tsx
T
DottaandPaperclip a6306ba606 feat(runner): consolidate Cursor production integration (#15075)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - The native Runner keeps provider sessions under company authority,
approvals, budgets and durable recovery.
> - Cursor work was spread across candidate branches. The published
branch lacked later plan, permission and cleanup fixes.
> - Production also needs public installation and matching runtime
assets for local and Daytona execution.
> - This pull request consolidates Cursor onto current mainline recovery
behavior and completes that installation path.
> - The installed v11 release passed focused local and Daytona
qualification after the generic mode and lifecycle cleanup. The later
model-selection correction and current mainline merge produce v14
artifacts that need matching release qualification.
> - Cursor admission is enabled in source; publish only an artifact
combination with matching qualification. Native AskQuestion and complete
per-run dollar accounting remain excluded.

## Linked Issues or Issue Description

Refs: #14435, #14631, #14669, #14699, #14724.

This completes the Cursor implementation by @cryppadotta from combined
source `22c78242a4e0c2369fecf0c2dc4e7600fbad6706`. It preserves newer
mainline recovery, completion and warm-directory behavior. Pi and
Copilot remain gated.

## What Changed

- Generate named Rust and TypeScript ACPX release profiles from one
manifest. Share runtime pins with packaging and server verification.
Preserve vendor runtime versions; bind the updated ACPX patch to Cursor
profile v14 and reject stale generated declarations at build/typecheck.
- Remove ACPX model allowlists, including the former Codex and Pi
restrictions and the duplicate developer test-drive gate. Send any
explicit model ID unchanged to its provider and verify the effective
selection before prompting. The bundled ACPX package forwards unlisted
IDs, rejects mismatched acknowledgements, and restores the exact
selection after session load. It does not expand Cursor model aliases.
Provider rejection, mismatch, or missing model controls fails without a
fallback. Model examples live in evaluation fixtures, outside runtime
declarations.

- Add pinned Cursor execution, contained instructions, exact model
verification and Agent/Plan/Ask modes.
- Carry an opaque generic `mode` identifier in shared native execution,
sidecar, Rust and recovery contracts. The provider adapter owns
supported modes, defaults, native translation and acknowledgement.
- Keep native RPC recognition, accepted-plan interpretation and
permission evidence behind provider adapters. Shared settlement and
recovery verify normalized facts and their committed evidence.
- Replace the Cursor-only warm-attachment branch with a runner-owned
capability. Only Cursor opts into it. Move profile compatibility and
optional usage parsing into provider metadata and adapters.
- Write generic plan-wait receipts. Read exact historical Cursor
receipts through a separate compatibility decoder. Reject mixed formats
and preserve existing authority checks.
- Carry native plans, semantic questions, todos, child activity,
permission identities and partial usage diagnostics through the Runner.
- Preserve durable response delivery, cancellation, warm ownership and
process retirement.
- Finish accepted planning runs successfully. Keep their tasks open for
explicit direction. Acceptance does not start implementation.
- Ship `paperclipai runtime setup cursor` and its provisioner through
the public package. npm installation does not download Cursor. Setup
uses the OS account's closure-keyed cache so system-wide npm packages
can remain read-only. Run it as the Paperclip service account.
- Include Cursor in normal provider packs and Daytona images for macOS
ARM64/x64 and Linux x64.
- Reject stale release packs by source revision and current ACPX/Cursor
pins before assembly writes files. Verify current Cursor
version/profile/closure again at runtime.
- Ship all three daemon targets and the expected Linux image-pack
identity. A macOS controller uses its packaged Linux daemon for Daytona.
Image mismatches fail before provider launch.
- Use the vendored Runner boundary for installed readiness probes.
Verify the actual installed Cursor probe.
- Verify compiled public Daytona plugins and their release versions in
installed smokes.
- Record exact artifacts, the acceptance matrix, retained failures,
supported capabilities and rollback behavior in the [readiness
report](https://github.com/paperclipai/paperclip/blob/codex/cursor-production-readiness/doc/plans/2026-10-03-cursor-production-readiness.md).

## Verification

- Current head `f7ec5cc1f0e30c62a829c016ff2013a2a9d79037` merges
mainline `faa8e452c73bae5e044dd6379179a00106abb131`. It keeps Cursor
plan and cancellation guards alongside mainline historical-question
filtering. The evaluation catalog includes both Cursor and expanded
adapter accounting cases (683 total). Recursive typecheck, full build,
696 lifecycle/recovery tests, 45 fixture tests and fixture typecheck
passed. Current-head CI passed: 56 successful checks, one neutral and
four skipped. [Complete
CI](https://github.com/paperclipai/paperclip/actions/runs/37557996535).
The fresh Base Greptile review is 5/5 on this exact head, with 304 files
reviewed, zero new comments and zero unresolved threads. The user
authorized overriding the CODEOWNER review gate after checks passed; no
failing checks are overridden. Prior results below retain their own head
identities.
- Corrective head `3d2b168366258036f6b6a6fccb382c49138cc601` fixes the
post-merge Apex finding. Automatic-review and new-evidence
reconciliation preserve pending child results and recheck delivery under
the status lock before completing. Account repair now excludes unrelated
secret consumers and requires the failed agent's identity. Regression
coverage includes the commit race, delivery statuses,
current-run/current-intent exclusions, repeated reconciliation, both
database reconciliation paths, and credential consumer boundaries. All
184 affected tests, server typecheck and server build passed.
Current-head Base Greptile review is 5/5, with 304 files reviewed, zero
new comments and zero unresolved threads. Current-head CI passed: 56
successful checks, one neutral and four skipped. [Complete
CI](https://github.com/paperclipai/paperclip/actions/runs/37535994724).
This Base review is distinct from the earlier Apex review.
- Merge head `5957c257a` reconciles mainline `b508a05c4`. It preserves
both accepted-plan waits and pending-child-completion checks, current
provider selectors, task-creation response identities, and mainline ACPX
missing-file handling. The combined patch is bound to Cursor profile
v14; historical records keep their original identities.
- Merge head `5957c257a` passed recursive typecheck, full build, 43
installed ACPX/package contracts, 107 provider UI and plan/recovery
tests, 593 database-backed lifecycle tests, 49 profile/native contract
tests, 45 Product E2E fixture tests, fixture typecheck, token gates,
three provider-free browser task-creation cases, and Runner
conformance/replay checks. Its complete CI passed (55 successful checks,
one neutral and four skipped), while Apex returned 2/5 with a
child-delivery finding addressed below.
- The local full-suite attempt again failed the unchanged Git streaming
test (360-second timeout) and was stopped. The concurrent local Rust
attempt failed four unchanged Codex process/deadline tests; all four
passed serially without code changes in 7.29 seconds after removing the
competing test load. These failed commands are retained and are not
reported as full-suite passes; the fresh Linux CI runs are tracked
separately.
- The previous head `907bdb2a2778c7ffeb4a662a91460c9d1ddfc9c5` earned
Apex 5/5 with zero comments after fixing all three findings: per-user
install cache, stale release-pack rejection, and public Linux smoke
account/home handling. Its real built installer passed from read-only
public packages on macOS ARM64 and Linux x64. All 137 release-registry
checks and 64 ACPX package contracts passed. That review does not cover
this mainline reconciliation.
- Prior `beadd3654` passed the full CI matrix; its one unchanged chat
test failure and successful single retry remain in the [CI
history](https://github.com/paperclipai/paperclip/actions/runs/37521449327).
Historical results below remain attributed to their original builds.

- Fixture follow-up `dd59d7e82b103a88b7cbd7d2c38b612c0fbbff7a` removes
provider-specific model choices from generic offline ACPX tests. The
fake sidecar preserves the model and session identity selected at open
through suspension. Affected verification passed: 106 Rust tests and 73
TypeScript tests. This commit changes test code only; the
production-code checks below retain their recorded identities. Its CI
and Greptile review later passed; those results belong to that
historical head.
- Model-selection cleanup `9a070808b48960a41fdfd369ae0636b95af82459`:
252 focused Runner tests passed (six platform skips), covering all six
ACPX agents, native model acknowledgement, rejected selections,
installation integrity and recovery identity. The merged branch passed
recursive typecheck, full build, token gates, server admission (19
tests), and the Product E2E catalog (45 tests). The acceptance catalog
passed all four tests. The full Rust suite passed: 643 tests, 2 ignored.
It verifies sidecar acknowledgement of unlisted models and rejection of
model mismatches. The final commits only update Rust tests; production
sources match the verified build at
`65ec3279ac50185e3cda109b5cfd9b4f56105de0`. No new paid provider calls
were made.
- The merge preserves both Cursor and the new mainline public-MCP
fixture cases. Auto-merge remains disabled; the latest follow-up status
is recorded above. The local `pnpm test:run` attempt hit the unchanged
Git streaming test's 300-second timeout and was interrupted before
merging mainline. The broad Runner attempt found obsolete single-model
assertions plus three macOS fixture-path failures caused by a
`/private/tmp` override. The assertions are corrected; affected
TypeScript checks passed with the standard macOS temporary directory,
and the complete Rust suite passed. Neither interrupted command is a
full-suite pass.
- Earlier declaration-cleanup head `6f4a5e9e2` passed recursive
typecheck, build, Rust and focused tests. Its CI later exposed a test
expecting duplicated Grok digest literals. The current source fixes that
assertion to compare launcher bytes with the shared manifest. Historical
successes and failed attempts are retained; no new live provider
qualification is claimed.
- Previous head `e75fde6098b0ddd8cec765bfb6ecaeecb88a26a6` passed
complete CI (56 successful checks, one neutral, four skipped) and
Greptile 5/5. [Historical complete
CI](https://github.com/paperclipai/paperclip/actions/runs/37489112305).
Those results are not claimed for the cleanup head.
- Frozen live application: `d7b696f9b8f79095233e9e3d56d23e6a6018dd48`.
Public package version: `0.0.0-cursor-verify.3d0c9b7761c6`. The
declaration cleanup preserves release pins and does not relabel that
tested artifact as a build of the new source. Mainline through
`e34abee670` was reconciled while preserving accepted-plan waits,
provider-capacity handling, and both Cursor and public-MCP fixtures.
- Clean normal installation, explicit Cursor setup and daemon resolution
passed on macOS ARM64, macOS x64 under Rosetta, and Linux x64. npm
lifecycle hooks ran without silently downloading Cursor.
- Historical v11 live matrix: **18/18 passed with cleanup** (nine local,
nine Daytona) after the generic mode and lifecycle cleanup. The campaign
has 23 attempts; all five failures and their diagnoses remain recorded.
Exact case identities, hashes and limits are in the readiness report.
All provider calls are real, use the explicit Luna model and
company-bound credentials, and run without qualification or
runtime-asset overrides.
- The immutable Daytona image is
`ghcr.io/paperclipai/paperclip-daytona-runner@sha256:d6259b6bba094702c13fc2283bd85550849c1c53145b656fb2746778f9fa1747`.
The public Daytona plugin is installed independently and its version is
checked.
- Recursive typecheck, full build, token gates and Runner
contract/conformance/replay checks passed on the frozen application. Its
complete Linux CI suite passed. The duplicate local full-suite command
was incomplete after timing failures; affected repeats passed, but that
command is not reported as a clean pass.
- Qualification fixtures passed typecheck, 1,675 Vitest tests (one
skip), 128 Node checks, three provider-free browser tests, and 150
focused lifecycle tests after the final diagnostic correction. The
affected legacy Cursor command file also passed all five tests after
removing its shorter 10-second override; it now inherits the suite’s
standard 15-second timeout. Greptile is 5/5 on `e75fde609` with no
unresolved review threads. CI results above are recorded separately from
historical build results.

## Risks

- Cursor v14 includes the updated ACPX dependency patch and release
identity. The v11 live matrix and image below remain historical
evidence. They do not certify new v14 package/image artifacts.

- ACPX accepts models beyond the qualification fixtures. Availability
and entitlement depend on the provider. Successful configuration is not
a claim of live qualification for every model.
- Shared mode is an opaque identifier. Provider adapters own its
meaning. Incompatible historical sessions remain fenced; exact committed
plan waits and task history remain inspectable.
- Native AskQuestion is excluded. Paperclip semantic questions are
supported. Authoritative per-run dollar accounting is unavailable;
partial counters remain diagnostics and unknown cost is not zero.
- Image input, detailed native diffs, deeper child transcripts and
native plan-file export remain follow-ups.
- macOS x64 has clean-install and daemon-startup proof under Rosetta,
not a separate live campaign on Intel hardware.
- Release only the tested package/image combination. Merging this PR
does not publish npm packages or deploy that image. Later builds need
their own release verification. Rollback disables new Cursor admission
while preserving records and recovery inspection.
- A model can fail an exact instruction: one cancelled-plan attempt
returned the wrong summary marker despite correct cancellation. The
unchanged repeat passed; both results remain in the report.

> ROADMAP.md was checked. This completes existing native Runner/Cursor
work; it does not add an independent core feature proposal.

## Model Used

OpenAI Codex, GPT-6. The exact serving variant and context window are
not exposed in this session. The agent used reasoning, repository
inspection, code execution, protocol tests and browser-backed Product
E2E tools. Cursor acceptance uses the explicit
`gpt-5.6-luna[context=272k,reasoning=medium,fast=false]` model. That is
the evaluated provider model.

## 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 — affected suites passed;
full CI and the retained local failed attempts are recorded separately
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 — 56 successful checks, one
neutral and four skipped on `f7ec5cc1f0e30c62a829c016ff2013a2a9d79037`
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups —
fresh Base review passed on `f7ec5cc1f0e30c62a829c016ff2013a2a9d79037`;
zero new comments and no unresolved threads. The earlier Apex finding
remains fixed.
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-10-06 20:48:15 -05:00

1590 lines
63 KiB
TypeScript

import { useCallback, useEffect, useLayoutEffect, useMemo, useRef, useState } from "react";
import {
CAPABILITY_DEFAULT_FIXTURE_PROFILE,
capabilityIssueThreadFixture,
} from "../../../src/issue-thread/fixtures";
import type {
CapabilityEvidenceSectionId,
CapabilityIssueThreadSnapshot,
} from "../../../src/issue-thread/types";
import type { CapabilityDevtoolsSnapshot } from "../../../src/devtools";
import { capabilityDenialCount } from "../../../src/issue-thread/types";
import { Composer } from "./Composer";
import { EvidencePanel } from "./EvidencePanel";
import type { CapabilityDevtoolsTab } from "./DevtoolsInspector";
import { Icon } from "./Icons";
import { IssueHeader } from "./IssueHeader";
import { applyFakeInteractionResponse } from "./fake-store";
import type { CapabilityInteractionResponse } from "./InteractionCard";
import {
capabilityLiveClient,
recallSession,
rememberSession,
type CapabilityCleanRoomIdentity,
type CapabilityHarnessConfiguration,
} from "./live-client";
import { parseCapabilityRoute, capabilityRouteHref, type CapabilityRoute } from "./route";
import { SurfaceNav, type CapabilityChatHistoryItem } from "./SurfaceNav";
import { EvalAssertions, TurnGroup, type EvalAssertion } from "./ThreadItems";
const PANEL_OPEN_KEY = "paperclip-runner.capability.panel.open";
/**
* The clean room keeps its own panel preference. Evidence is collapsed by
* default there on purpose (revision 5: "the default view reads as plain
* chat"), and inheriting an explorer session's opened drawer would quietly
* break that on the very first visit.
*/
const CHAT_PANEL_OPEN_KEY = "paperclip-runner.capability.chat.panel.open";
const PANEL_WIDTH_KEY = "paperclip-runner.capability.panel.width";
const PANEL_MIN = 320;
const PANEL_MAX = 960;
/** Play-all cadence for the replay strip (§6); slow enough to read a turn. */
const REPLAY_STEP_MS = 800;
/**
* The pending-request guidance is a constant because the settle path has to
* recognise its own stale announcement: a screen reader must not still be told
* a request is waiting after it was answered (TASK-16978).
*/
const PENDING_ANNOUNCEMENT = "A request is waiting for your answer.";
const ANSWERED_ANNOUNCEMENT = "Your answer was recorded.";
const SETTLED_ANNOUNCEMENT = "The pending request is resolved.";
const SCENARIOS = ["hb-baseline", "dp-documents", "ix-interactions", "ar-artifacts", "wc-workspace-changes", "fr-file-reference", "pe-structured-plan", "te-command-execution", "mp-mcp-progress", "rs-web-research", "da-subagent-delegation", "mr-model-routing", "cc-context-compaction", "ag-generated-artifact", "rv-review-mode", "hk-hook-lifecycle", "mc-memory-citations", "sr-safety-review", "ti-terminal-input", "wt-intentional-wait", "pn-provider-notices", "ax-acpx-lifecycle", "ax-acpx-permissions", "ax-acpx-tools", "ax-acpx-events", "ax-acpx-failures"];
const CHAT_HISTORY_KEY = "paperclip-runner.capability.chat.history.v1";
const CHAT_HARNESS_KEY = "paperclip-runner.capability.chat.harness.v1";
const MODEL_PRESETS = {
codex: ["gpt-5.4-mini", "gpt-5.4"],
opencode: ["openrouter/deepseek/deepseek-v4-flash-0731"],
acpx: ["openrouter/deepseek/deepseek-v4-flash-0731", "claude-sonnet-5", "gpt-5.6-sol"],
} as const;
const ACPX_DEFAULT_MODELS = {
claude: "claude-sonnet-5",
codex: "gpt-5.6-sol",
} as const;
function acpxLabel(configuration: CapabilityHarnessConfiguration | null | undefined): string {
if (configuration?.provider !== "acpx") return configuration?.provider ?? "starting";
const agent = configuration.acpxAgent ?? "codex";
return `Real ${agent === "claude" ? "Claude" : "Codex"} via ACPX`;
}
interface EmbeddedEvalCheck {
id: string;
title: string;
description: string;
passed: boolean;
detail: string;
definition: Record<string, unknown>;
anchor: { kind: "item" | "turn" | "run"; id: string };
}
interface EmbeddedEvalReport {
publication?: { schema: string; notice: string };
attemptId: string;
caseId: string;
disposition: string;
passed: boolean;
checks: EmbeddedEvalCheck[];
run: {
model: string;
provider: string;
driver: string;
providerVersion: string | null;
runnerProvider: string;
configuration: string;
sessionId: string;
providerSessionId: string | null;
agentVersion: string | null;
retainedSession: boolean | null;
retainedSessionStatus: string | null;
fixtureDigest: string;
runnerPackageDigest: string;
runnerdDigest: string;
runnerBuild: string;
startedAt: string;
finishedAt: string;
durationMs: number | null;
initialRevision: number | null;
finalRevision: number | null;
finalStateSummary?: string;
usage: {
agentTurns: number;
providerRequests: number | null;
inputTokens?: number;
outputTokens?: number;
cachedInputTokens?: number;
reasoningTokens?: number;
providerReportedCostNanodollars?: number;
estimatedCostNanodollars?: number;
pricingVersion: string;
} | null;
};
view: CapabilityIssueThreadSnapshot;
devtools: CapabilityDevtoolsSnapshot | null;
navigation: { suiteHref: string; previous: { label: string; href: string } | null; next: { label: string; href: string } | null };
}
declare global {
interface Window { __PAPERCLIP_EVAL_REPORT__?: EmbeddedEvalReport }
}
interface StoredChatSession {
sessionId: string;
snapshot: CapabilityIssueThreadSnapshot;
identity: CapabilityCleanRoomIdentity | null;
updatedAt: string;
configuration?: CapabilityHarnessConfiguration;
runtime?: Awaited<ReturnType<typeof capabilityLiveClient.newCleanRoom>>["runtime"];
}
function defaultHarness(provider: CapabilityHarnessConfiguration["provider"] = "codex"): CapabilityHarnessConfiguration {
return {
provider,
model: MODEL_PRESETS[provider][0],
...(provider === "acpx" ? { acpxAgent: "codex" as const } : {}),
lifecyclePolicy: { mode: "warm", idleTimeoutMs: 300_000 },
};
}
function readHarnessConfiguration(): CapabilityHarnessConfiguration {
try {
const value = JSON.parse(window.localStorage.getItem(CHAT_HARNESS_KEY) ?? "null") as Partial<CapabilityHarnessConfiguration> | null;
if (value && (value.provider === "codex" || value.provider === "opencode" || value.provider === "acpx") && typeof value.model === "string") {
const lifecyclePolicy = value.lifecyclePolicy?.mode === "per_turn"
? { mode: "per_turn" as const, idleTimeoutMs: null }
: value.lifecyclePolicy?.mode === "warm" && Number.isSafeInteger(value.lifecyclePolicy.idleTimeoutMs) && Number(value.lifecyclePolicy.idleTimeoutMs) > 0
? { mode: "warm" as const, idleTimeoutMs: Number(value.lifecyclePolicy.idleTimeoutMs) }
: { mode: "warm" as const, idleTimeoutMs: 300_000 };
return {
provider: value.provider,
model: value.model,
...(value.provider === "acpx" ? {
acpxAgent: value.acpxAgent === "claude" ? value.acpxAgent : "codex",
} : {}),
lifecyclePolicy,
};
}
} catch {
// Invalid preferences fall back to the qualified defaults.
}
return defaultHarness();
}
function persistHarnessConfiguration(configuration: CapabilityHarnessConfiguration): void {
try {
window.localStorage.setItem(CHAT_HARNESS_KEY, JSON.stringify(configuration));
} catch {
// Provider/model preference is convenient but not authority-bearing.
}
}
function readChatHistory(): StoredChatSession[] {
try {
const parsed = JSON.parse(window.localStorage.getItem(CHAT_HISTORY_KEY) ?? "[]") as unknown;
if (!Array.isArray(parsed)) return [];
return parsed.filter((item): item is StoredChatSession =>
typeof item === "object" && item !== null &&
typeof (item as StoredChatSession).sessionId === "string" &&
typeof (item as StoredChatSession).updatedAt === "string" &&
typeof (item as StoredChatSession).snapshot === "object",
).slice(0, 12);
} catch {
return [];
}
}
function persistChatHistory(history: StoredChatSession[]): void {
try {
window.localStorage.setItem(CHAT_HISTORY_KEY, JSON.stringify(history.slice(0, 12)));
} catch {
// History is a convenience; a blocked or full storage quota must not stop chat.
}
}
function readStoredNumber(key: string, fallback: number): number {
try {
const raw = window.localStorage.getItem(key);
const parsed = raw === null ? Number.NaN : Number.parseInt(raw, 10);
return Number.isFinite(parsed) ? Math.min(PANEL_MAX, Math.max(PANEL_MIN, parsed)) : fallback;
} catch {
return fallback;
}
}
function readStoredFlag(key: string, fallback: boolean): boolean {
try {
const raw = window.localStorage.getItem(key);
return raw === null ? fallback : raw === "true";
} catch {
return fallback;
}
}
/**
* Scroll an Evidence target into the panel viewport. `start` alignment has to
* account for the sticky panel head, which would otherwise sit on top of the
* section header the deep link was supposed to reveal.
*/
function scrollEvidenceIntoView(target: Element, block: "start" | "center"): void {
target.scrollIntoView({ block });
if (block !== "start") return;
const panel = target.closest<HTMLElement>(".pit-panel");
const head = panel?.querySelector<HTMLElement>(".pit-panel-chrome");
if (panel == null || head == null) return;
panel.scrollTop = Math.max(0, panel.scrollTop - head.offsetHeight);
}
function describe(cause: unknown): string {
return String(cause instanceof Error ? cause.message : cause);
}
/** Resolution copy for a request that settled without this client answering it. */
function settledAnnouncement(snapshot: CapabilityIssueThreadSnapshot, interactionId: string): string {
for (const turn of snapshot.turns) {
for (const item of turn.items) {
if (item.kind === "interaction" && item.interactionId === interactionId) {
return `${SETTLED_ANNOUNCEMENT} ${item.stateLabel}.`;
}
}
}
return SETTLED_ANNOUNCEMENT;
}
function currentTurnActivity(snapshot: CapabilityIssueThreadSnapshot): string {
const harness = snapshot.identity.agentLabel.replace(/^Real /, "");
const turn = snapshot.turns.at(-1);
if (turn === undefined) return `Dispatching turn to ${harness}`;
const activity = [...turn.items].reverse().find((item) =>
item.kind === "tool_activity" || item.kind === "progress_activity" || item.kind === "provider_activity" ||
(item.kind === "agent_message" && item.streaming),
);
if (activity?.kind === "tool_activity") {
return activity.status === "running"
? `Paperclip tool · ${activity.operationId}`
: `Paperclip tool completed · ${activity.operationId}`;
}
if (activity?.kind === "progress_activity") return activity.summary;
if (activity?.kind === "agent_message") return `Receiving ${harness} response`;
return `Waiting for ${harness} activity`;
}
function useRoute(): CapabilityRoute {
const [route, setRoute] = useState(() => parseCapabilityRoute(window.location));
useEffect(() => {
const onChange = () => setRoute(parseCapabilityRoute(window.location));
window.addEventListener("hashchange", onChange);
window.addEventListener("popstate", onChange);
return () => {
window.removeEventListener("hashchange", onChange);
window.removeEventListener("popstate", onChange);
};
}, []);
return route;
}
function useLayout(): "side" | "overlay" | "segment" {
const [layout, setLayout] = useState<"side" | "overlay" | "segment">(() =>
window.innerWidth <= 767 ? "segment" : window.innerWidth <= 1100 ? "overlay" : "side",
);
useEffect(() => {
const onResize = () =>
setLayout(window.innerWidth <= 767 ? "segment" : window.innerWidth <= 1100 ? "overlay" : "side");
window.addEventListener("resize", onResize);
return () => window.removeEventListener("resize", onResize);
}, []);
return layout;
}
export function App() {
const embeddedEval = window.__PAPERCLIP_EVAL_REPORT__ ?? null;
const route = useRoute();
const chat = route.surface === "chat";
const layout = useLayout();
useEffect(() => {
if (embeddedEval !== null) {
document.title = `${embeddedEval.passed ? "✓" : "✕"} ${embeddedEval.caseId} · paperclip-runner eval`;
return;
}
document.title = chat
? "🫧 Mock Paperclip · Issue thread"
: "🧯 Mock Paperclip · Issue thread";
}, [chat, embeddedEval]);
const [snapshot, setSnapshot] = useState<CapabilityIssueThreadSnapshot | null>(null);
const [devtools, setDevtools] = useState<CapabilityDevtoolsSnapshot | null>(null);
const [identity, setIdentity] = useState<CapabilityCleanRoomIdentity | null>(null);
const initialHarnessRef = useRef<CapabilityHarnessConfiguration>(readHarnessConfiguration());
const [harness, setHarness] = useState<CapabilityHarnessConfiguration>(initialHarnessRef.current);
const [activeHarness, setActiveHarness] = useState<CapabilityHarnessConfiguration | null>(null);
const [providerRuntime, setProviderRuntime] = useState<Awaited<ReturnType<typeof capabilityLiveClient.newCleanRoom>>["runtime"]>(undefined);
const [chatHistory, setChatHistory] = useState<StoredChatSession[]>(readChatHistory);
const [historicSessionId, setHistoricSessionId] = useState<string | null>(null);
const liveRoomRef = useRef<StoredChatSession | null>(null);
/**
* Which surface produced `snapshot`. A hash change commits the new route in
* the same frame that still renders the old snapshot, so without this a
* capture or a test can catch one frame of "settled" that belongs to the
* surface it just navigated away from.
*/
const [snapshotSurface, setSnapshotSurface] = useState(route.surface);
const [error, setError] = useState<string | null>(null);
const [actionError, setActionError] = useState<string | null>(null);
const [loadNonce, setLoadNonce] = useState(0);
const [settled, setSettled] = useState(false);
const [panelOpen, setPanelOpen] = useState(() => embeddedEval !== null ||
readStoredFlag(route.surface === "chat" ? CHAT_PANEL_OPEN_KEY : PANEL_OPEN_KEY, false));
const [panelWidth, setPanelWidth] = useState(() => readStoredNumber(PANEL_WIDTH_KEY, 384));
const [devtoolsTab, setDevtoolsTab] = useState<CapabilityDevtoolsTab>(embeddedEval !== null ? "eval" : "evidence");
const [segment, setSegment] = useState<"thread" | "evidence">(route.segment);
const [openSections, setOpenSections] = useState<CapabilityEvidenceSectionId[]>(["tools"]);
const [selectedTurnId, setSelectedTurnId] = useState<string | "all">("all");
const [highlightedRecordId, setHighlightedRecordId] = useState<string | null>(route.record);
const [focusInteractionId, setFocusInteractionId] = useState<string | null>(null);
const [confirmReset, setConfirmReset] = useState(false);
const [announcement, setAnnouncement] = useState("");
const [showJump, setShowJump] = useState(false);
const [reconnected, setReconnected] = useState(false);
const [playing, setPlaying] = useState(false);
/** True while a turn stream is open, so the surface never reads as settled. */
const [streamingTurn, setStreamingTurn] = useState(false);
const scrollRef = useRef<HTMLDivElement | null>(null);
const cancelResetRef = useRef<HTMLButtonElement | null>(null);
/** Last announced pending request, so the live region tracks transitions. */
const announcedPendingRef = useRef<string | null>(null);
/**
* Turn generation. Reset, `New chat`, and leaving the surface all bump it, so
* a frame from a turn that has since been abandoned is dropped instead of
* appended to a thread it no longer belongs to.
*/
const turnGenerationRef = useRef(0);
const turnAbortRef = useRef<AbortController | null>(null);
useEffect(() => {
if (embeddedEval !== null) setPanelOpen(true);
}, [embeddedEval]);
/** Abandons any open turn stream: the server sees the disconnect and stops. */
const abandonTurn = useCallback(() => {
turnGenerationRef.current += 1;
const controller = turnAbortRef.current;
turnAbortRef.current = null;
setStreamingTurn(false);
controller?.abort();
}, []);
/* --------------------------------------------------------------- loading */
useEffect(() => {
let cancelled = false;
setSettled(false);
setError(null);
if (embeddedEval !== null) {
setIdentity(null);
setSnapshot({
...embeddedEval.view,
composer: { state: "disabled", helper: null, reason: "Immutable eval recording", pendingInteractionId: null },
});
setDevtools(embeddedEval.devtools);
setSnapshotSurface("issue");
} else if (chat) {
// The clean room has no fixture fallback: if real Codex and real runnerd
// cannot start, the surface says so rather than rendering a canned thread.
void (async () => {
try {
const response = await capabilityLiveClient.loadCleanRoom(
recallSession("cleanroom"),
initialHarnessRef.current,
);
if (cancelled) return;
rememberSession(response.sessionId, "cleanroom");
setIdentity(response.identity ?? null);
const configuration = response.configuration ?? initialHarnessRef.current;
setActiveHarness(configuration);
// The server-returned immutable session configuration is authoritative
// for both the active status and the options initially shown for the
// next chat after a restore.
setHarness(configuration);
setProviderRuntime(response.runtime);
setHistoricSessionId(null);
setSnapshot(response.view);
setSnapshotSurface("chat");
} catch (cause) {
if (!cancelled) setError(String(cause instanceof Error ? cause.message : cause));
}
})();
} else if (route.mode === "live") {
void (async () => {
try {
const response = await capabilityLiveClient.load(recallSession());
if (cancelled) return;
rememberSession(response.sessionId);
setSnapshot(response.view);
setSnapshotSurface("issue");
} catch (cause) {
if (!cancelled) setError(String(cause instanceof Error ? cause.message : cause));
}
})();
} else {
setIdentity(null);
setSnapshot(capabilityIssueThreadFixture(route.shot ?? "thread-baseline", route.fixtureProfile));
setSnapshotSurface("issue");
}
return () => {
cancelled = true;
// Leaving the surface (or reloading it) must not leave a turn streaming
// into a thread that is no longer on screen.
abandonTurn();
};
}, [abandonTurn, chat, embeddedEval, route.mode, route.shot, route.fixtureProfile, loadNonce]);
/* ------------------------------------------------------- deep-link params */
useEffect(() => {
if (route.panel === null) return;
setPanelOpen(true);
setSegment("evidence");
setOpenSections((current) =>
current.includes(route.panel as CapabilityEvidenceSectionId)
? current
: [...current, route.panel as CapabilityEvidenceSectionId],
);
setHighlightedRecordId(route.record);
}, [route.panel, route.record]);
useLayoutEffect(() => {
// A deep link has to land on what it addressed. With `rec` that is the
// record; without it, the opened section's header — otherwise the link
// half-arrives with the section still below the fold. Wait for the bundled
// faces before measuring: a host fallback can otherwise move the retained
// scroll position by a pixel even after the final webfonts replace it.
if (route.panel === null || snapshot === null) return;
const panel = route.panel;
const record = route.record;
let cancelled = false;
void (async () => {
if (typeof document.fonts?.ready?.then === "function") {
await document.fonts.ready;
}
if (cancelled) return;
const target =
record === null
? document.querySelector(`[data-evidence-section="${CSS.escape(panel)}"]`)
: document.querySelector(`[data-record-id="${CSS.escape(record)}"]`);
if (target !== null) {
scrollEvidenceIntoView(target, record === null ? "start" : "center");
}
})();
return () => {
cancelled = true;
};
}, [route.panel, route.record, snapshot, panelOpen, segment, openSections]);
useEffect(() => {
setSegment(route.segment);
}, [route.segment]);
/* -------------------------------------------------- settle + auto-follow */
useEffect(() => {
if (snapshot === null) return;
let cancelled = false;
const scroller = scrollRef.current;
void (async () => {
if (typeof document.fonts?.ready?.then === "function") {
await document.fonts.ready;
}
if (cancelled) return;
if (scroller !== null) scroller.scrollTop = embeddedEval === null ? scroller.scrollHeight : 0;
await new Promise<void>((resolve) => requestAnimationFrame(() => resolve()));
if (!cancelled) setSettled(true);
})();
return () => {
cancelled = true;
};
}, [snapshot]);
useEffect(() => {
if (embeddedEval !== null || !panelOpen || snapshot === null || route.mode !== "live") return;
if (historicSessionId !== null) return;
let cancelled = false;
void capabilityLiveClient.devtools(snapshot.sessionId)
.then((next) => { if (!cancelled) setDevtools(next); })
.catch((cause) => { if (!cancelled) setActionError(describe(cause)); });
return () => { cancelled = true; };
}, [embeddedEval, historicSessionId, panelOpen, route.mode, snapshot?.renderedAt, snapshot?.sessionId]);
useEffect(() => {
if (!chat || snapshot === null || historicSessionId !== null) return;
const record: StoredChatSession = {
sessionId: snapshot.sessionId,
snapshot,
identity,
updatedAt: snapshot.renderedAt,
...(activeHarness === null ? {} : { configuration: activeHarness }),
...(providerRuntime === undefined ? {} : { runtime: providerRuntime }),
};
liveRoomRef.current = record;
setChatHistory((current) => {
const next = [record, ...current.filter((item) => item.sessionId !== record.sessionId)].slice(0, 12);
persistChatHistory(next);
return next;
});
}, [activeHarness, chat, historicSessionId, identity, providerRuntime, snapshot?.renderedAt, snapshot?.sessionId]);
useEffect(() => {
if (snapshot === null) return;
const pending = snapshot.composer.pendingInteractionId;
const announced = announcedPendingRef.current;
announcedPendingRef.current = pending;
if (pending !== null) {
// Announce a request once, when it arrives. Re-announcing on every
// snapshot would talk over the outcome the user just triggered: a live
// submit re-renders with the same request still pending.
if (pending !== announced) setAnnouncement(PENDING_ANNOUNCEMENT);
return;
}
if (snapshot.connection.state === "reconnecting") {
setAnnouncement(`Connection lost — retrying (attempt ${snapshot.connection.attempt}).`);
return;
}
if (announced === null) return;
// The request settled. Anything the resolution itself announced stands;
// only the now-false "still waiting" guidance is replaced, so a screen
// reader never carries pending-state copy into later turns (TASK-16978).
setAnnouncement((current) =>
current === PENDING_ANNOUNCEMENT ? settledAnnouncement(snapshot, announced) : current,
);
}, [snapshot]);
/* -------------------------------------------------------------- handlers */
const persistPanel = useCallback(
(open: boolean, width: number) => {
try {
window.localStorage.setItem(chat ? CHAT_PANEL_OPEN_KEY : PANEL_OPEN_KEY, String(open));
window.localStorage.setItem(PANEL_WIDTH_KEY, String(width));
} catch {
// Panel preference is a convenience, never a correctness requirement.
}
},
[chat],
);
const openEvidence = useCallback(
(section: CapabilityEvidenceSectionId, recordId: string) => {
setPanelOpen(true);
setDevtoolsTab("evidence");
setSegment("evidence");
setSelectedTurnId("all");
setOpenSections((current) => (current.includes(section) ? current : [...current, section]));
setHighlightedRecordId(recordId);
persistPanel(true, panelWidth);
window.setTimeout(() => {
document
.querySelector(`[data-record-id="${CSS.escape(recordId)}"]`)
?.scrollIntoView({ block: "center" });
}, 0);
},
[panelWidth, persistPanel],
);
const closeEvidence = useCallback(() => {
setPanelOpen(false);
persistPanel(false, panelWidth);
if (layout === "segment") setSegment("thread");
// Focus came into the panel when it opened (§9.2), so hand it back to the
// visible control that owns the panel rather than dropping it on `body`.
window.setTimeout(() => {
const candidates = ['[data-testid="evidence-toggle"]', '[data-testid="segment-thread"]'];
for (const selector of candidates) {
const control = document.querySelector<HTMLElement>(selector);
if (control !== null && control.offsetParent !== null) {
control.focus();
return;
}
}
}, 0);
}, [layout, panelWidth, persistPanel]);
const jumpToThread = useCallback((anchorId: string) => {
setSegment("thread");
const anchor = document.getElementById(anchorId) ?? document.querySelector(`#${CSS.escape(anchorId)}`);
anchor?.scrollIntoView({ block: "center" });
}, []);
const respond = useCallback(
(response: CapabilityInteractionResponse) => {
if (snapshot === null) return;
if (route.mode !== "live") {
setSnapshot(applyFakeInteractionResponse(snapshot, response));
setAnnouncement(ANSWERED_ANNOUNCEMENT);
return;
}
// Network effects must never live inside a React state updater. Strict
// Mode intentionally invokes updater functions more than once in
// development, which previously submitted every interaction twice: the
// first request resolved it and the duplicate produced "not pending".
const sessionId = snapshot.sessionId;
setActionError(null);
setSnapshot((current) => current === null ? current : {
...current,
turns: current.turns.map((turn) => ({
...turn,
items: turn.items.map((item) =>
item.kind === "interaction" && item.interactionId === response.interactionId
? { ...item, state: "submitting" as const, stateLabel: "Submitting…" }
: item,
),
})),
});
void capabilityLiveClient
.respond(sessionId, response.interactionId, response.outcome, response.result)
.then((next) => {
setSnapshot(next.view);
setAnnouncement(ANSWERED_ANNOUNCEMENT);
})
.catch((cause) => {
setActionError(describe(cause));
setSnapshot((current) => current === null ? current : {
...current,
turns: current.turns.map((turn) => ({
...turn,
items: turn.items.map((item) =>
item.kind === "interaction" &&
item.interactionId === response.interactionId &&
item.state === "submitting"
? { ...item, state: "pending" as const, stateLabel: "Waiting for you" }
: item,
),
})),
});
});
},
[route.mode, snapshot],
);
/**
* Sends one message and renders the turn as it arrives.
*
* Each frame is a whole server projection, so the surface still never patches
* state locally — it just gets more than one projection per turn. The settled
* payload is applied last and stays the authority.
*/
const send = useCallback(
(message: string) => {
if (snapshot === null) return;
if (route.mode !== "live") return;
setActionError(null);
abandonTurn();
const generation = turnGenerationRef.current;
const controller = new AbortController();
turnAbortRef.current = controller;
setStreamingTurn(true);
setSnapshot((current) =>
current === null ? current : { ...current, composer: { ...current.composer, state: "sending" } },
);
void capabilityLiveClient
.send(snapshot.sessionId, message, {
signal: controller.signal,
onFrame: (view) => {
if (turnGenerationRef.current === generation) setSnapshot(view);
},
})
.then((next) => {
if (turnGenerationRef.current === generation) {
setSnapshot(next.view);
setProviderRuntime(next.runtime);
}
})
.catch((cause) => {
if (turnGenerationRef.current !== generation || controller.signal.aborted) return;
setActionError(describe(cause));
// A refused turn must not strand the composer in `sending`.
setSnapshot((stale) =>
stale === null ? stale : { ...stale, composer: { ...stale.composer, state: "ready" } },
);
})
.finally(() => {
if (turnGenerationRef.current !== generation) return;
turnAbortRef.current = null;
setStreamingTurn(false);
});
},
[abandonTurn, route.mode, snapshot],
);
const stop = useCallback(() => {
setSnapshot((current) => {
if (current === null) return current;
if (route.mode === "live") {
// Stop reaches the provider interrupt through its own request; the open
// turn stream keeps rendering and its terminal frame — not this
// response — decides what the stopped turn finally looks like.
const streaming = turnAbortRef.current !== null;
void capabilityLiveClient
.stop(current.sessionId)
.then((next) => {
if (!streaming && turnAbortRef.current === null) setSnapshot(next.view);
})
.catch((cause) => setActionError(describe(cause)));
return current;
}
const turns = current.turns.map((turn, index) =>
index === current.turns.length - 1 ? { ...turn, stoppedByUser: true } : turn,
);
return {
...current,
turns,
composer: { state: "ready", helper: null, reason: null, pendingInteractionId: null },
};
});
setAnnouncement("Turn stopped. Partial output is preserved.");
}, [route.mode]);
/**
* Adopts a rotated clean room. Reset and `New chat` both land here: the
* server always answers with a new session id and new mock identities, so the
* client's job is only to forget the old ones.
*/
const adoptCleanRoom = useCallback((next: Awaited<ReturnType<typeof capabilityLiveClient.newCleanRoom>>) => {
rememberSession(next.sessionId, "cleanroom");
setIdentity(next.identity ?? null);
const configuration = next.configuration ?? harness;
setActiveHarness(configuration);
setHarness(configuration);
setProviderRuntime(next.runtime);
setHistoricSessionId(null);
setSnapshot(next.view);
setDevtools(null);
setError(null);
setActionError(null);
setAnnouncement(
`New clean-room chat started on ${next.view.issue.identifier}. The previous session was closed.`,
);
}, [harness]);
const newChat = useCallback(() => {
const model = harness.model?.trim() ?? "";
if (!model) {
setActionError("Choose or enter a model before starting a new chat.");
return;
}
if (harness.provider === "opencode" && !model.includes("/")) {
setActionError("OpenCode models must use provider/model form.");
return;
}
const configuration = { ...harness, model };
persistHarnessConfiguration(configuration);
setConfirmReset(false);
// The room this turn belongs to is about to be retired, so the stream is
// dropped before the request that retires it.
abandonTurn();
setAnnouncement("Starting a new clean-room chat…");
void capabilityLiveClient
.newCleanRoom(liveRoomRef.current?.sessionId ?? snapshot?.sessionId ?? null, configuration)
.then(adoptCleanRoom)
.catch((cause) => setActionError(describe(cause)));
}, [abandonTurn, adoptCleanRoom, harness, snapshot]);
const selectChatHistory = useCallback((sessionId: string) => {
const live = liveRoomRef.current;
if (live?.sessionId === sessionId) {
setSnapshot(live.snapshot);
setIdentity(live.identity);
setActiveHarness(live.configuration ?? null);
if (live.configuration !== undefined) setHarness(live.configuration);
setProviderRuntime(live.runtime);
setDevtools(null);
setHistoricSessionId(null);
setActionError(null);
return;
}
const archived = chatHistory.find((item) => item.sessionId === sessionId);
if (archived === undefined) return;
abandonTurn();
const configuration = archived.configuration ?? defaultHarness();
void capabilityLiveClient.loadCleanRoom(sessionId, configuration)
.then((restored) => {
rememberSession(restored.sessionId, "cleanroom");
setSnapshot(restored.view);
setIdentity(restored.identity ?? archived.identity);
setActiveHarness(restored.configuration ?? configuration);
setHarness(restored.configuration ?? configuration);
setProviderRuntime(restored.runtime);
setDevtools(null);
setHistoricSessionId(null);
setActionError(null);
setAnnouncement("Archived chat restored. The runner will resume when you send a message.");
})
.catch((cause) => setActionError(describe(cause)));
}, [abandonTurn, chatHistory]);
const reset = useCallback(() => {
setConfirmReset(false);
abandonTurn();
if (chat && snapshot !== null) {
void capabilityLiveClient
.reset(snapshot.sessionId)
.then(adoptCleanRoom)
.catch((cause) => setActionError(describe(cause)));
return;
}
if (route.mode === "live" && snapshot !== null) {
void capabilityLiveClient
.reset(snapshot.sessionId)
.then((next) => {
rememberSession(next.sessionId);
setSnapshot(next.view);
})
.catch((cause) => setActionError(describe(cause)));
return;
}
setSnapshot(capabilityIssueThreadFixture("thread-baseline", route.fixtureProfile));
setAnnouncement("Scenario reset. The mock state is back to its clean seed.");
}, [abandonTurn, adoptCleanRoom, chat, route.fixtureProfile, route.mode, snapshot]);
const retry = useCallback(() => {
if (route.mode === "live" && snapshot !== null) {
void capabilityLiveClient
.reconnect(snapshot.sessionId)
.then((next) => setSnapshot(next.view))
.catch((cause) => setActionError(describe(cause)));
return;
}
setSnapshot((current) =>
current === null
? current
: {
...current,
connection: { state: "connected", attempt: 0 },
composer: { state: "ready", helper: null, reason: null, pendingInteractionId: null },
},
);
setReconnected(true);
window.setTimeout(() => setReconnected(false), 3_000);
}, [route.mode, snapshot]);
useEffect(() => {
if (!confirmReset) return;
cancelResetRef.current?.focus();
function onKeyDown(event: KeyboardEvent) {
if (event.key === "Escape") setConfirmReset(false);
}
document.addEventListener("keydown", onKeyDown);
return () => document.removeEventListener("keydown", onKeyDown);
}, [confirmReset]);
/* ----------------------------------------------------------- replay (§6) */
// `?at=<ordinal>` is the source of truth for where the recording is parked,
// so step / next-turn / play-all and the deep link all move the same value.
const replay = useMemo(() => {
if (snapshot === null || snapshot.replay === null) return null;
const { total } = snapshot.replay;
const ordinal =
route.at === null ? snapshot.replay.ordinal : Math.min(total, Math.max(0, route.at));
return { ordinal, total };
}, [snapshot, route.at]);
const seekReplay = useCallback(
(ordinal: number) => {
setPlaying(false);
window.location.hash = capabilityRouteHref(route, { at: ordinal });
},
[route],
);
useEffect(() => {
if (!playing || replay === null) return;
if (replay.ordinal >= replay.total) {
setPlaying(false);
return;
}
const timer = window.setTimeout(() => {
window.location.hash = capabilityRouteHref(route, { at: replay.ordinal + 1 });
}, REPLAY_STEP_MS);
return () => window.clearTimeout(timer);
}, [playing, replay, route]);
useEffect(() => {
if (replay === null) setPlaying(false);
}, [replay]);
/* ---------------------------------------------------------------- render */
const denialCount = useMemo(
() => (snapshot === null ? 0 : capabilityDenialCount(snapshot.evidence, null)),
[snapshot],
);
const historyItems: CapabilityChatHistoryItem[] = chatHistory.map((item) => ({
sessionId: item.sessionId,
identifier: item.snapshot.issue.identifier,
title: item.snapshot.issue.title,
updatedAt: item.updatedAt,
current: liveRoomRef.current?.sessionId === item.sessionId,
}));
const surfaceNav = (
<SurfaceNav
surface={route.surface}
history={historyItems}
activeSessionId={snapshot?.sessionId ?? null}
onNewChat={chat ? newChat : () => { window.location.hash = "#/chat"; }}
onSelectHistory={selectChatHistory}
/>
);
if (error !== null) {
return (
<div className="pit-app" data-thread-state="failed" data-surface={route.surface}>
{surfaceNav}
<main className="pit-app-error">
<p className="pit-composer-reason" role="alert" data-testid="surface-error">
{chat
? `The clean-room chat could not start the selected provider: ${error}`
: error}
</p>
{chat ? (
<>
<p className="pit-muted">
The clean room only runs against the selected real provider through real runnerd, so it does not fall
back to a fixture or a recording.
</p>
<section className="pit-harness-picker" aria-label="Chat provider recovery options">
<label>
<span>Provider</span>
<select
data-testid="chat-provider"
value={harness.provider}
onChange={(event) => {
const provider = event.target.value as CapabilityHarnessConfiguration["provider"];
setHarness((current) => ({
...defaultHarness(provider),
lifecyclePolicy: current.lifecyclePolicy,
}));
}}
>
<option value="codex">Codex</option>
<option value="opencode">OpenCode</option>
<option value="acpx">ACPX</option>
</select>
</label>
{harness.provider === "acpx" ? (
<label>
<span>ACP agent</span>
<select
data-testid="chat-acpx-agent"
value={harness.acpxAgent ?? "codex"}
onChange={(event) => {
const acpxAgent = event.target.value as NonNullable<CapabilityHarnessConfiguration["acpxAgent"]>;
setHarness((current) => ({
...current,
acpxAgent,
model: ACPX_DEFAULT_MODELS[acpxAgent],
}));
}}
>
<option value="claude">Claude</option>
<option value="codex">Codex (control)</option>
</select>
</label>
) : null}
<label>
<span>Execution</span>
<select
data-testid="chat-lifecycle-mode"
value={harness.lifecyclePolicy.mode}
onChange={(event) => setHarness((current) => ({
...current,
lifecyclePolicy: event.target.value === "per_turn"
? { mode: "per_turn", idleTimeoutMs: null }
: { mode: "warm", idleTimeoutMs: 300_000 },
}))}
>
<option value="warm">Warm session</option>
<option value="per_turn">Turn by turn</option>
</select>
</label>
{harness.lifecyclePolicy.mode === "warm" ? (
<label>
<span>Idle timeout (seconds)</span>
<input
data-testid="chat-idle-timeout"
type="number"
min={1}
value={Math.round((harness.lifecyclePolicy.idleTimeoutMs ?? 300_000) / 1_000)}
onChange={(event) => setHarness((current) => ({
...current,
lifecyclePolicy: {
mode: "warm",
idleTimeoutMs: Math.max(1, Number(event.target.value)) * 1_000,
},
}))}
/>
</label>
) : null}
<label className="pit-harness-model">
<span>Model</span>
<input
data-testid="chat-model"
list={`chat-recovery-models-${harness.provider}`}
value={harness.model ?? ""}
placeholder={harness.provider === "opencode" ? "provider/model" : "model name"}
onChange={(event) => setHarness((current) => ({ ...current, model: event.target.value }))}
/>
<datalist id={`chat-recovery-models-${harness.provider}`}>
{MODEL_PRESETS[harness.provider].map((model) => <option key={model} value={model} />)}
</datalist>
</label>
<button
type="button"
className="pit-button"
data-variant="primary"
data-testid="chat-apply-harness"
onClick={newChat}
>
Start new chat
</button>
</section>
</>
) : null}
<button
type="button"
className="pit-button"
data-variant="primary"
data-testid="surface-error-retry"
onClick={() => setLoadNonce((current) => current + 1)}
>
Try again
</button>
</main>
</div>
);
}
if (snapshot === null) {
return (
<div className="pit-app" data-thread-state="loading" data-surface={route.surface}>
{surfaceNav}
<main className="pit-app-error">
<p className="pit-muted" role="status" data-testid="surface-loading">
{chat
? "Starting real runnerd and the selected provider for a fresh mock tenant…"
: "Loading…"}
</p>
</main>
</div>
);
}
const showThread = layout !== "segment" || segment === "thread";
const showPanel = layout === "segment" ? segment === "evidence" : panelOpen;
return (
<div
className="pit-app"
data-thread-state={
streamingTurn
? "streaming"
: settled && snapshotSurface === route.surface
? "settled"
: "loading"
}
data-session-mode={snapshot.mode}
data-surface={route.surface}
data-connection-state={snapshot.connection.state}
data-eval-view={embeddedEval !== null}
>
{embeddedEval === null ? surfaceNav : null}
<IssueHeader
snapshot={snapshot}
readOnly={embeddedEval !== null}
scenarios={SCENARIOS}
surface={route.surface}
cleanRoomToken={identity?.token ?? null}
evidenceOpen={panelOpen}
denialCount={denialCount}
segment={segment}
onToggleEvidence={() => {
const next = !panelOpen;
setPanelOpen(next);
persistPanel(next, panelWidth);
if (layout === "segment") setSegment(next ? "evidence" : "thread");
}}
onSelectScenario={(scenario) => {
window.location.hash = capabilityRouteHref(route, { fixtureProfile: scenario });
}}
onReplay={() => {
window.location.hash = capabilityRouteHref(route, {
shot: "replay-mode",
mode: "replay",
at: 12,
});
}}
onReset={() => setConfirmReset(true)}
onStop={stop}
onSelectSegment={setSegment}
/>
{chat && embeddedEval === null ? (
<section className="pit-harness-picker" aria-label="Chat provider and model">
<label>
<span>Provider</span>
<select
data-testid="chat-provider"
value={harness.provider}
disabled={streamingTurn}
onChange={(event) => {
const provider = event.target.value as CapabilityHarnessConfiguration["provider"];
setHarness((current) => ({
...defaultHarness(provider),
lifecyclePolicy: current.lifecyclePolicy,
}));
}}
>
<option value="codex">Codex</option>
<option value="opencode">OpenCode</option>
<option value="acpx">ACPX</option>
</select>
</label>
{harness.provider === "acpx" ? (
<label>
<span>ACP agent</span>
<select
data-testid="chat-acpx-agent"
value={harness.acpxAgent ?? "codex"}
disabled={streamingTurn}
onChange={(event) => {
const acpxAgent = event.target.value as NonNullable<CapabilityHarnessConfiguration["acpxAgent"]>;
setHarness((current) => ({
...current,
acpxAgent,
model: ACPX_DEFAULT_MODELS[acpxAgent],
}));
}}
>
<option value="claude">Claude</option>
<option value="codex">Codex (control)</option>
</select>
</label>
) : null}
<label>
<span>Execution</span>
<select
data-testid="chat-lifecycle-mode"
value={harness.lifecyclePolicy.mode}
disabled={streamingTurn}
onChange={(event) => setHarness((current) => ({
...current,
lifecyclePolicy: event.target.value === "per_turn"
? { mode: "per_turn", idleTimeoutMs: null }
: { mode: "warm", idleTimeoutMs: 300_000 },
}))}
>
<option value="warm">Warm session</option>
<option value="per_turn">Turn by turn</option>
</select>
</label>
{harness.lifecyclePolicy.mode === "warm" ? (
<label>
<span>Idle timeout (seconds)</span>
<input
data-testid="chat-idle-timeout"
type="number"
min={1}
step={1}
value={Math.round(harness.lifecyclePolicy.idleTimeoutMs / 1_000)}
disabled={streamingTurn}
onChange={(event) => {
const seconds = Math.max(1, Number.parseInt(event.target.value || "1", 10));
setHarness((current) => ({
...current,
lifecyclePolicy: { mode: "warm", idleTimeoutMs: seconds * 1_000 },
}));
}}
/>
</label>
) : null}
<label className="pit-harness-model">
<span>Model</span>
<input
data-testid="chat-model"
list={`chat-models-${harness.provider}`}
value={harness.model ?? ""}
disabled={streamingTurn}
placeholder={harness.provider === "opencode" ? "provider/model" : "model name"}
onChange={(event) => setHarness((current) => ({ ...current, model: event.target.value }))}
/>
<datalist id={`chat-models-${harness.provider}`}>
{MODEL_PRESETS[harness.provider].map((model) => <option key={model} value={model} />)}
</datalist>
</label>
<button
type="button"
className="pit-button"
data-variant="primary"
data-testid="chat-apply-harness"
disabled={streamingTurn || historicSessionId !== null}
onClick={newChat}
>
Start new chat
</button>
<span className="pit-harness-active" data-testid="chat-active-harness">
Active: {acpxLabel(activeHarness)}
{activeHarness?.model ? ` · ${activeHarness.model}` : ""}
{activeHarness ? ` · ${activeHarness.lifecyclePolicy.mode === "warm" ? `warm ${Math.round(activeHarness.lifecyclePolicy.idleTimeoutMs / 1_000)}s` : "turn by turn"}` : ""}
{providerRuntime?.runnerPid ? ` · runner PID ${providerRuntime.runnerPid}` : ""}
{activeHarness?.provider === "acpx" && providerRuntime?.sidecarPid ? ` · sidecar PID ${providerRuntime.sidecarPid}` : ""}
{activeHarness?.provider === "acpx" && providerRuntime?.agentPid ? ` · agent PID ${providerRuntime.agentPid}` : ""}
{activeHarness?.provider === "acpx" && providerRuntime?.driverSessionId ? ` · ACPX record ${providerRuntime.driverSessionId}` : ""}
{providerRuntime?.providerSessionId ? ` · session ${providerRuntime.providerSessionId}` : ""}
{activeHarness?.provider === "acpx" && providerRuntime?.providerVersion ? ` · ACPX ${providerRuntime.providerVersion}` : ""}
{activeHarness?.provider === "acpx" && providerRuntime?.agentServerVersion ? ` · agent ${providerRuntime.agentServerVersion}` : ""}
{activeHarness?.provider === "acpx" && providerRuntime?.acpProtocolVersion ? ` · ACP ${providerRuntime.acpProtocolVersion}` : ""}
{providerRuntime?.status ? ` · ${providerRuntime.status}` : ""}
</span>
</section>
) : null}
{embeddedEval !== null ? (
<nav className="pit-eval-nav" aria-label="Eval result navigation">
<a href={embeddedEval.navigation.suiteHref}>← All results</a>
{embeddedEval.navigation.previous ? <a href={embeddedEval.navigation.previous.href}>← {embeddedEval.navigation.previous.label}</a> : <span />}
<strong>{embeddedEval.passed ? "PASS" : "FAIL"} · {embeddedEval.attemptId}</strong>
{embeddedEval.navigation.next ? <a href={embeddedEval.navigation.next.href}>{embeddedEval.navigation.next.label} →</a> : <span />}
</nav>
) : null}
{actionError !== null ? (
<p className="pit-banner" data-tone="danger" role="alert" data-testid="action-error">
<span aria-hidden="true">⚠</span>
{actionError}
<button
type="button"
className="pit-link-button"
onClick={() => setActionError(null)}
>
Dismiss
</button>
</p>
) : null}
{snapshot.connection.state === "reconnecting" ? (
<p className="pit-banner" role="status" data-testid="reconnect-banner">
<span aria-hidden="true">⏳</span>
Connection lost — retrying (attempt {snapshot.connection.attempt})
</p>
) : reconnected ? (
<p className="pit-banner" data-tone="success" role="status">
<span aria-hidden="true">✓</span>
Reconnected
</p>
) : null}
{streamingTurn ? (
<div className="pit-live-activity" role="status" aria-live="polite" data-testid="live-activity">
<span className="pit-live-activity-pulse" aria-hidden="true" />
<Icon name="terminal" />
<span>{currentTurnActivity(snapshot)}</span>
<span className="pit-live-activity-tail" aria-hidden="true">live</span>
</div>
) : null}
{replay !== null ? (
<div className="pit-replay-strip" data-testid="replay-strip">
<span>
Replay {replay.ordinal}/{replay.total}
</span>
<div
className="pit-progress"
role="progressbar"
aria-valuemin={0}
aria-valuemax={replay.total}
aria-valuenow={replay.ordinal}
aria-label="Replay progress"
>
<div
className="pit-progress-fill"
style={{ width: `${(replay.ordinal / replay.total) * 100}%` }}
/>
</div>
<button
type="button"
className="pit-button"
disabled={replay.ordinal <= 0}
data-testid="replay-step-back"
onClick={() => seekReplay(replay.ordinal - 1)}
>
Step back
</button>
<button
type="button"
className="pit-button"
disabled={replay.ordinal >= replay.total}
data-testid="replay-next-turn"
onClick={() => seekReplay(replay.ordinal + 1)}
>
Next turn
</button>
<button
type="button"
className="pit-button"
data-variant={playing ? undefined : "primary"}
aria-pressed={playing}
disabled={!playing && replay.ordinal >= replay.total}
data-testid="replay-play-all"
onClick={() => setPlaying((current) => !current)}
>
<span aria-hidden="true">{playing ? "❚❚" : "▶"}</span>
{playing ? "Pause" : "Play all"}
</button>
</div>
) : null}
<div className="pit-body">
<main className="pit-main" hidden={!showThread}>
<div className="pit-main-inner">
<div
className="pit-thread-scroll"
ref={scrollRef}
tabIndex={0}
aria-label="Conversation"
onScroll={(event) => {
const element = event.currentTarget;
const distance =
element.scrollHeight - element.scrollTop - element.clientHeight;
setShowJump(distance > 300);
}}
>
<div className="pit-thread">
{embeddedEval !== null ? (
<div className="pit-eval-boundary" data-phase="execution"><strong>Eval execution</strong>{embeddedEval.publication ? <span>{embeddedEval.publication.notice}</span> : null}</div>
) : null}
{snapshot.turns.length === 0 ? (
<section className="pit-empty-thread" data-testid="clean-room-empty">
<h2>Start a clean-room chat</h2>
<p>
This is a blank thread on a brand-new mock tenant:{" "}
<strong>{snapshot.issue.identifier}</strong> in{" "}
<strong>Mock Paperclip (clean room)</strong>. Nothing has been said, called, or
recorded yet.
</p>
<p className="pit-muted">
Your first message starts a real {snapshot.identity.agentLabel.replace(/^Real /, "")} turn through real runnerd. The agent may
call the semantic tools this session exposes, and every record it creates lands
in the mock control plane only — never a real Paperclip API. Detailed tool,
policy, event, and state evidence stays in the Evidence drawer until you open
it.
</p>
</section>
) : (
snapshot.turns.map((turn) => (
<TurnGroup
key={turn.id}
turn={turn}
callbacks={{
onOpenEvidence: openEvidence,
onRespond: respond,
focusInteractionId,
}}
assertions={Object.fromEntries(embeddedEval?.checks
.filter((check) => check.anchor.kind === "item")
.map((check) => [check.anchor.id, [{ ...check } satisfies EvalAssertion]]) ?? [])}
terminalAssertions={embeddedEval?.checks
.filter((check) => check.anchor.kind === "turn" && check.anchor.id === turn.id)
.map((check) => ({ ...check })) ?? []}
/>
))
)}
{embeddedEval !== null ? (
<div className="pit-eval-boundary" data-phase="post-run">
<strong>Post-run state</strong>
<span>{embeddedEval.publication ? "Company-state details withheld from public replay" : embeddedEval.run.finalStateSummary ?? `Final mock control-plane revision ${embeddedEval.run.finalRevision}`}</span>
<EvalAssertions assertions={embeddedEval.checks.filter((check) => check.anchor.kind === "run")} />
</div>
) : null}
</div>
</div>
{showJump ? (
<button
type="button"
className="pit-button pit-jump-pill"
onClick={() => {
const element = scrollRef.current;
if (element !== null) element.scrollTop = element.scrollHeight;
setShowJump(false);
}}
>
Jump to latest
</button>
) : null}
</div>
{embeddedEval === null ? <Composer
model={snapshot.composer}
sessionId={snapshot.sessionId}
onSend={send}
onStop={stop}
onRetry={retry}
onReset={() => (chat ? newChat() : setConfirmReset(true))}
resetLabel={chat ? "New chat" : "Reset scenario"}
onFocusPending={(interactionId) => {
setFocusInteractionId(interactionId);
document
.getElementById(`interaction-${interactionId}`)
?.scrollIntoView({ block: "center" });
}}
/> : null}
</main>
{showPanel && layout === "side" ? (
<div
className="pit-splitter"
role="separator"
aria-orientation="vertical"
aria-label="Resize the evidence panel"
aria-valuemin={PANEL_MIN}
aria-valuemax={PANEL_MAX}
aria-valuenow={panelWidth}
tabIndex={0}
onPointerDown={(event) => {
const splitter = event.currentTarget;
splitter.setPointerCapture(event.pointerId);
const resize = (pointer: PointerEvent) => {
const next = Math.min(PANEL_MAX, Math.max(PANEL_MIN, window.innerWidth - pointer.clientX));
setPanelWidth(next);
};
const finish = () => {
splitter.removeEventListener("pointermove", resize);
splitter.removeEventListener("pointerup", finish);
splitter.removeEventListener("pointercancel", finish);
setPanelWidth((current) => {
persistPanel(panelOpen, current);
return current;
});
};
splitter.addEventListener("pointermove", resize);
splitter.addEventListener("pointerup", finish);
splitter.addEventListener("pointercancel", finish);
}}
onKeyDown={(event) => {
const step = event.shiftKey ? 64 : 16;
if (event.key === "ArrowLeft") {
event.preventDefault();
setPanelWidth((current) => {
const next = Math.min(PANEL_MAX, current + step);
persistPanel(panelOpen, next);
return next;
});
} else if (event.key === "ArrowRight") {
event.preventDefault();
setPanelWidth((current) => {
const next = Math.max(PANEL_MIN, current - step);
persistPanel(panelOpen, next);
return next;
});
}
}}
/>
) : null}
{showPanel ? (
<EvidencePanel
snapshot={snapshot}
devtoolsTab={devtoolsTab}
onDevtoolsTabChange={setDevtoolsTab}
evalReport={embeddedEval}
{...(embeddedEval !== null || (route.mode === "live" && historicSessionId === null) ? { devtools } : {})}
onForkRevision={(revision) => {
abandonTurn();
setActionError(null);
void capabilityLiveClient.fork(snapshot.sessionId, revision)
.then((next) => {
rememberSession(next.sessionId, chat ? "cleanroom" : "issue");
setSnapshot(next.view);
setIdentity(next.identity ?? null);
setDevtools(null);
})
.catch((cause) => setActionError(describe(cause)));
}}
layout={layout}
width={panelWidth}
selectedTurnId={selectedTurnId}
openSections={openSections}
highlightedRecordId={highlightedRecordId}
onSelectTurn={setSelectedTurnId}
onToggleSection={(section) =>
setOpenSections((current) =>
current.includes(section)
? current.filter((entry) => entry !== section)
: [...current, section],
)
}
onClose={closeEvidence}
onJumpToThread={jumpToThread}
onInvokeTool={route.mode === "live" && historicSessionId === null ? async (operationId, input) => {
setActionError(null);
const next = await capabilityLiveClient.invokeTool(snapshot.sessionId, operationId, input);
setSnapshot(next.view);
setSelectedTurnId(next.toolTurnId);
return next.toolResult;
} : undefined}
/>
) : null}
</div>
{confirmReset ? (
<div className="pit-dialog-backdrop">
<div
className="pit-dialog"
role="dialog"
aria-modal="true"
aria-labelledby="reset-dialog-title"
data-testid="reset-dialog"
>
<h2 className="pit-dialog-title" id="reset-dialog-title">
{chat ? "Reset this chat?" : "Reset scenario?"}
</h2>
<p>
{chat
? "This stops any active turn, closes this session's authority, and opens a new mock tenant with new identities. The transcript and its mock records will be lost."
: "This clears the mock state and starts a clean session. The transcript will be lost."}
</p>
<div className="pit-button-row">
<button
type="button"
className="pit-button"
ref={cancelResetRef}
onClick={() => setConfirmReset(false)}
>
Cancel
</button>
<button
type="button"
className="pit-button"
data-variant="destructive"
onClick={reset}
data-testid="reset-confirm"
>
{chat ? "Reset chat" : "Reset scenario"}
</button>
</div>
</div>
</div>
) : null}
<p
className="pit-visually-hidden"
role="status"
aria-live="polite"
data-testid="thread-live-region"
>
{announcement}
</p>
</div>
);
}