mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-15 01:52:55 +01:00
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The chat-feature evals measure whether agents complete real chat requests through the product > - The chat catalog eval harness and its repairs live on the `chat-eval/master-candidate` branch, not on master > - Today's work made 139 of 242 chat evals pass in verified single-case live runs. Each product fix landed on master as its own pull request > - The eval-side changes (setups, graders, fault seams, evidence capture) still exist only on the candidate > - This pull request lands the harness on master. Master stays the source of truth for product behavior > - The benefit is one tree: evals run against master commits, and product fixes reach the evals as soon as they merge ## Linked Issues or Issue Description Refs #19441, Refs #19599, Refs #20270, Refs #20334, Refs #20342, Refs #20344, Refs #20354, Refs #20360, Refs #20373, Refs #20382, Refs #20409 **What existing behavior does this improve?** The chat catalog eval harness and the product it evaluates live on a long-running branch outside master. Eval fixes do not reach master, so master-based eval runs use old setups and graders. **Subsystem affected** `tests/runner-e2e` chat catalog harness, plus a small set of product changes that exist only on the candidate. **Proposed change** Land the harness on master. Every product fix from this eval work already merged to master as its own pull request. Conflicts keep master's product behavior. ## What Changed - Most of the diff (808 of 851 files) is under `tests/runner-e2e/`: chat-catalog domains, scenarios, descriptors, the TestOps manifest, private evidence tooling, and their tests. - Graders now check saved state, setup literals and rendered or clicked links. Many keyword and regex checks on free-form replies are removed. Each changed grader has a bumped `oracleVersion` and calibration tests (a valid reply, a reworded valid reply, and broken variants that must fail). - The shared link helper `hasLink` no longer treats a list link (for example `/COM/issues?view=mine`) as a link to every task without a URL key. - Setup fixes build real state through supported product routes: routine history, agent grants through the board grant route, budget holds with a real cap, and a storage-layer archive fault for `inbox-search.archive-partial-failure`. - Browser boundary journeys (sign-in, chat-disabled, onboarding) now write the standard pass evidence files. - Retargeted cases, approved by the maintainer: `agent-chat.chat-disabled` (Conference Room gate), `agents.restore-instructions` (no saved instruction history), `access.manage-membership` (member permission edits stay board-only). - The private evidence trusted files (`manual-private-handoff.ts`, `manual-private-retention.ts`, `manual-private-report-assets.ts`) are byte-identical to master. - Product files outside `tests/` (43 files, +1,328 −49) carry candidate-only changes that master does not yet have. Most are tests. The main code changes are in `request-human-input.ts`, `heartbeat-run-summary.ts`, `native-completion-feedback.ts`, `native-run-finalizer.ts`, `openapi.ts`, `secrets.ts` and `IssueChatLongThreadPerf.tsx`. ## Verification - `pnpm vitest run chat-catalog-domains chat-catalog-contract.test.ts chat-catalog-consistency.test.ts unit-launcher.test.mjs --testTimeout=180000` in `tests/runner-e2e`: 269 files pass, 3,550 tests pass. - `node --test chat-full-catalog-testops-manifest.test.mjs`: 10 pass. - `node tests/runner-e2e/source-sdk-public-inputs.mjs <HEAD> --check`: no change. - `tsc` for `server` and `tests/runner-e2e`: clean. - Live single-case verification: 139 of 242 cases passed on paid runs, and a person read the saved evidence for each. These are single-case runs, not one full-cohort run. ## Risks - Large diff. Most of it is test-only, but the 43 non-test files change product code paths. Review those files first. - Some graders are now less strict on reply wording. A person must read replies for meaning on those cases. - 42 cases need disposable outside accounts (Notion, Slack, GitHub, Daytona and others) and stay blocked. `access.invite-teammate` stays failing until local agent isolation lands, because the runner's durable state keeps a live invite token. - The paid workflow compares the trusted private-evidence files with master. They are unchanged, so this risk is low. ## Model Used Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use and code execution, in Claude Code. Subagents used the same 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 - [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 - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Paperclip <noreply@paperclip.ing>
165 lines
19 KiB
TypeScript
165 lines
19 KiB
TypeScript
import { chatCatalogNativePhaseDeadline, chatCatalogNativePhaseIsTimely, chatCatalogNativePhaseRemainingBudget } from "../chat-catalog-contract.js";
|
|
import { createHash } from "node:crypto";
|
|
import { readFile, realpath } from "node:fs/promises";
|
|
import path from "node:path";
|
|
import { expect } from "@playwright/test";
|
|
import type { FeedbackTrace, FeedbackTraceBundle, FeedbackVote } from "../../../packages/shared/src/types/feedback.js";
|
|
import { pollUntil } from "../api.js";
|
|
import { sendChatMessage } from "../chat-flow.js";
|
|
import { checks, type Entity } from "./internal-support.js";
|
|
import { prepareRuntimeWorkerGate, readRuntimeTaskGate, releaseRuntimeWorker } from "./runtime-busy-worker.js";
|
|
import { probeAgentPermission, cleanupIdentityProbes } from "./identity-support.js";
|
|
import { prepareOwnedCli, runOwnedCli, cleanupOwnedCli } from "./privileged-cli-support.js";
|
|
import { privilegedDefinition, privilegedObservation, companionPhase } from "./privileged-support.js";
|
|
const pendingFixtures = new WeakMap<Parameters<typeof prepareRuntimeWorkerGate>[0], { gate: Awaited<ReturnType<typeof prepareRuntimeWorkerGate>>; taskId?: string }>();
|
|
const sha = (b: Buffer | string) => createHash("sha256").update(b).digest("hex");
|
|
/** CLI exports use stored ZIP entries. Independently compare every local entry,
|
|
* reject hidden extras and unsafe names, and scan every actual archive byte. */
|
|
export function storedArchiveEntries(bytes: Buffer): Map<string, Buffer> {
|
|
const rows = new Map<string, Buffer>(); let offset = 0;
|
|
while (offset + 4 <= bytes.length && bytes.readUInt32LE(offset) === 0x04034b50) {
|
|
if (offset + 30 > bytes.length || bytes.readUInt16LE(offset + 6) & 9 || bytes.readUInt16LE(offset + 8) !== 0) throw new Error("Unsupported feedback archive entry");
|
|
const compressed = bytes.readUInt32LE(offset + 18), size = bytes.readUInt32LE(offset + 22), nameSize = bytes.readUInt16LE(offset + 26), extraSize = bytes.readUInt16LE(offset + 28);
|
|
const begin = offset + 30 + nameSize + extraSize, end = begin + compressed;
|
|
const name = bytes.subarray(offset + 30, offset + 30 + nameSize).toString("utf8");
|
|
if (compressed !== size || end > bytes.length || !name || name.startsWith("/") || name.split("/").includes("..") || name.includes("\\") || rows.has(name)) throw new Error("Unsafe or inconsistent feedback archive entry");
|
|
rows.set(name, bytes.subarray(begin, end)); offset = end;
|
|
}
|
|
if (!rows.size || offset + 4 > bytes.length || bytes.readUInt32LE(offset) !== 0x02014b50) throw new Error("Missing feedback ZIP directory");
|
|
return rows;
|
|
}
|
|
export function interruptedBundleMatches(bundle: FeedbackTraceBundle, ids: { company: string; issue: string; run: string; agent: string; comment: string }) {
|
|
const run = bundle.paperclipRun;
|
|
const runFile = bundle.files.find(f => f.path === "paperclip/run.json"), eventsFile = bundle.files.find(f => f.path === "paperclip/run-events.json");
|
|
if (!runFile || !eventsFile) return false;
|
|
try {
|
|
const saved = JSON.parse(runFile.contents) as Record<string, unknown>, events = JSON.parse(eventsFile.contents) as Array<{ runId?: string; agentId?: string; seq?: number }>;
|
|
return bundle.companyId === ids.company && bundle.issueId === ids.issue && bundle.envelope.targetId === ids.comment
|
|
&& run?.id === ids.run && run.companyId === ids.company && run.agentId === ids.agent && run.status === "cancelled"
|
|
&& saved.id === ids.run && saved.status === "cancelled" && saved.agentId === ids.agent && saved.companyId === ids.company
|
|
&& events.length > 0 && events.every(e => e.runId === ids.run) && Number(run.eventCount) === events.length;
|
|
} catch { return false; }
|
|
}
|
|
export function gradeFeedbackBundle(value: unknown) {
|
|
const c = checks(), o = privilegedObservation(value); c.add("actual-local-feedback-export", !!o, "A real human-owned scoped CLI export and retained interrupted-work trace are required."); if (!o) return c.result();
|
|
c.add("board-only-report-boundary", o.state.agentReportStatus === 403 && o.state.cliExecuted === true, "The actual agent cannot read feedback traces; an isolated human CLI actor performs the export.");
|
|
c.add("interrupted-work-not-fabricated-failure", o.state.workerWasReallyWaiting === true && o.state.cancelledBeforeCanonical === true
|
|
&& o.state.exportedInterruptionMatches === true && o.runs.some(r => r.id === o.state.workerRunId && r.status === "cancelled")
|
|
&& o.state.originalTraceUnchanged === true, "The original actual cancelled run, source-attributed comment and exact retained events survive export; cancellation is never relabeled as a failed native/model outcome.");
|
|
c.add("owned-local-bytes-and-scope", o.state.bundleWithinOwnedDirectory === true && o.state.exportCompanyMatches === true && o.state.exportIssueMatches === true
|
|
&& o.state.fullTraceBytesMatch === true && o.state.archiveBytesMatch === true && typeof o.state.zipSha256 === "string" && Number(o.state.zipBytes) > 0
|
|
&& o.state.noPrivateCredentialBytes === true && o.state.remoteSharingUnchanged === true && o.state.bundlePublished === false,
|
|
"Actual JSON and archive bytes match the intended company/issue trace only, remain private and do not enable remote sharing.");
|
|
const report = o.state.exportReport as { attributed?: boolean; reply?: string } | undefined;
|
|
// The saved state above proves scope, bytes and sharing. The reply must be the
|
|
// attributed answer to the receipt and give the real local location; a person
|
|
// reads it for meaning (wording checks rejected "has not been published").
|
|
c.add("truthful-native-location-and-contents", report?.attributed === true && report.reply?.includes(String(o.state.localOutputDirectory)) === true,
|
|
"The actual follow-up answers the operator's export receipt and reports the real local location.");
|
|
return c.result();
|
|
}
|
|
const feedback = privilegedDefinition("cli-operations.feedback-bundle", "privileged-feedback", {
|
|
async setup(ctx) {
|
|
const gate = await prepareRuntimeWorkerGate(ctx, { taskBound: true }); const pending = { gate, taskId: undefined as string | undefined }; pendingFixtures.set(ctx, pending);
|
|
const marker = `Waiting for operator brief ${ctx.input.nonce}`;
|
|
const task = await ctx.input.api.post<Entity>(`${gate.company}/issues`, { title: "Interrupted local feedback fixture", status: "todo", assigneeAgentId: gate.worker.id,
|
|
description: `First post one normal task comment saying exactly ${JSON.stringify(marker)}. Then run ${gate.gate.command} once and wait for the human brief. Do not mark this task complete or replace the script. This is an owned interruption fixture.` });
|
|
pending.taskId = task.id;
|
|
const running = await pollUntil({ label: "Actual worker comment and native waiting process", deadlineAt: Date.now() + 150000,
|
|
load: async () => ({ runs: await ctx.allRuns(), comments: await ctx.input.api.get<Entity[]>(`/api/issues/${task.id}/comments?order=asc`), waiting: await readRuntimeTaskGate(ctx, gate.gate, task.id, gate.worker.id) }),
|
|
accept: o => o.waiting === "waiting" && o.runs.length === 1 && o.runs[0]?.status === "running" && o.runs[0].agentId === gate.worker.id
|
|
&& o.comments.some(c => c.authorAgentId === gate.worker.id && c.createdByRunId === o.runs[0]!.id && c.body === marker),
|
|
reject: o => o.runs.some(r => ["failed", "cancelled", "timed_out", "succeeded"].includes(r.status)) ? "Interruption prerequisite finished before the actual held native boundary" : undefined });
|
|
const run = running.runs[0]!, comment = running.comments.find(c => c.body === marker && c.createdByRunId === run.id)!;
|
|
await ctx.input.api.post(`/api/heartbeat-runs/${run.id}/cancel`, {});
|
|
await pollUntil({ label: "Actual board cancellation before feedback request", deadlineAt: Date.now() + 60000, load: () => ctx.allRuns(), accept: runs => runs.length === 1 && runs[0]?.status === "cancelled" });
|
|
await ctx.input.api.patch(`/api/issues/${task.id}`, { status: "backlog" });
|
|
await releaseRuntimeWorker({ gate: gate.gate });
|
|
const vote = await ctx.input.api.post<FeedbackVote>(`/api/issues/${task.id}/feedback-votes`, { targetType: "issue_comment", targetId: comment.id,
|
|
vote: "down", reason: "The operator interrupted this unfinished fixture before completion; retain the original cancelled run and events for local review.", allowSharing: false });
|
|
const trace = await pollUntil({ label: "Actual human feedback trace retained locally", deadlineAt: Date.now() + 30000,
|
|
load: () => ctx.input.api.get<FeedbackTrace[]>(`/api/issues/${task.id}/feedback-traces`), accept: rows => rows.length === 1 && rows[0]?.feedbackVoteId === vote.id });
|
|
// The human's own vote answers the instance's "prompt" sharing preference
|
|
// (allowSharing: false saves not_allowed). Read the baseline after it so the
|
|
// export, not the fixture's vote, is what must leave sharing unchanged.
|
|
const sharingBefore = await ctx.input.api.get<{ feedbackDataSharingPreference?: string }>("/api/instance/settings/general");
|
|
if (sharingBefore.feedbackDataSharingPreference === "allowed") throw new Error("Feedback fixture vote enabled remote sharing");
|
|
const agentReportStatus = await probeAgentPermission(ctx, `/api/issues/${task.id}/feedback-traces`, "GET");
|
|
return { worker: gate.worker, gate: gate.gate, task, workerRunId: run.id, workerCommentId: comment.id, traceId: trace[0]!.id,
|
|
traceBefore: { id: trace[0]!.id, targetId: trace[0]!.targetId, status: trace[0]!.status, destination: trace[0]!.destination, exportId: trace[0]!.exportId },
|
|
sharingPreferenceBefore: sharingBefore.feedbackDataSharingPreference, workerWasReallyWaiting: true, cancelledBeforeCanonical: true, agentReportStatus, ...(await prepareOwnedCli(ctx)) };
|
|
},
|
|
async act(ctx, state) {
|
|
await pollUntil({ label: "Canonical human-export guidance settled", deadlineAt: chatCatalogNativePhaseDeadline(ctx, Date.now() + 150000),
|
|
load: async () => { await ctx.refreshIssue(); return { issue: ctx.issue(), runs: await ctx.allRuns() }; },
|
|
accept: o => chatCatalogNativePhaseIsTimely(ctx) && (o.issue?.conversationState === "waiting" && o.runs.length === 2 && o.runs.filter(r => r.status === "succeeded").length === 1 && o.runs.filter(r => r.status === "cancelled").length === 1) });
|
|
const prior = await ctx.allRuns(); const company = ctx.input.fixtures.company.id, task = state.task as Entity;
|
|
const output = path.join(String(state.operatorDirectory), "feedback-export");
|
|
const result = JSON.parse(await runOwnedCli(ctx, ["feedback", "export", "--company-id", company, "--issue-id", task.id, "--out", output, "--json"])); state.cliExecuted = true;
|
|
const directory = await realpath(output), root = await realpath(String(state.operatorDirectory));
|
|
state.bundleWithinOwnedDirectory = directory.startsWith(root + path.sep) && result.outputDir === output && result.zipPath === output + ".zip";
|
|
if (!state.bundleWithinOwnedDirectory) throw new Error("Feedback CLI output escapes its exact owned local destination");
|
|
const indexBytes = await readFile(path.join(directory, "index.json")), index = JSON.parse(indexBytes.toString()) as { companyId: string; summary: { total: number; issues: string[] }; files: { fullTraces: string[]; traces: string[]; votes: string[] } };
|
|
state.exportCompanyMatches = index.companyId === company; state.exportIssueMatches = index.summary.total === 1
|
|
&& index.summary.issues.length === 1 && index.summary.issues[0] === (task.identifier ?? task.id) && index.files.fullTraces.length === 1 && index.files.traces.length === 1 && index.files.votes.length === 1;
|
|
if (!state.exportIssueMatches) throw new Error("Feedback export contains unexpected company or issue scope");
|
|
const traceDir = index.files.fullTraces[0]!;
|
|
if (path.basename(traceDir) !== traceDir) throw new Error("Unexpected feedback trace directory");
|
|
const bundleBytes = await readFile(path.join(directory, "full-traces", traceDir, "bundle.json")), bundle = JSON.parse(bundleBytes.toString()) as FeedbackTraceBundle;
|
|
state.exportedInterruptionMatches = interruptedBundleMatches(bundle, { company, issue: task.id, run: String(state.workerRunId), agent: (state.worker as Entity).id, comment: String(state.workerCommentId) });
|
|
const actual = await ctx.input.api.get<FeedbackTraceBundle>(`/api/feedback-traces/${state.traceId}/bundle`);
|
|
state.fullTraceBytesMatch = JSON.stringify(bundle) === JSON.stringify(actual);
|
|
const zip = await readFile(result.zipPath), archive = storedArchiveEntries(zip);
|
|
const archivePrefix = path.basename(directory) + "/";
|
|
const expected = new Map<string, Buffer>([["index.json", indexBytes]]);
|
|
for (const group of ["votes", "traces"] as const) for (const name of index.files[group]) {
|
|
if (path.basename(name) !== name) throw new Error("Unexpected exported feedback filename");
|
|
expected.set(`${group}/${name}`, await readFile(path.join(directory, group, name)));
|
|
}
|
|
expected.set(`full-traces/${traceDir}/bundle.json`, bundleBytes);
|
|
for (const file of bundle.files) {
|
|
if (path.isAbsolute(file.path) || file.path.split(/[\\/]/).includes("..")) throw new Error("Unsafe local trace output path");
|
|
expected.set(`full-traces/${traceDir}/${file.path}`, Buffer.from(file.contents));
|
|
}
|
|
state.archiveBytesMatch = archive.size === expected.size && [...expected].every(([file, bytes]) => archive.get(archivePrefix + file)?.equals(bytes));
|
|
state.zipBytes = zip.length; state.zipSha256 = sha(zip);
|
|
state.noPrivateCredentialBytes = [indexBytes, bundleBytes, zip].every(bytes => !(ctx.input.privateSecrets ?? []).some(secret => secret && bytes.includes(Buffer.from(secret))));
|
|
if (!state.noPrivateCredentialBytes) throw new Error("Private credential bytes detected in local feedback output; matched content withheld");
|
|
const after = await ctx.input.api.get<FeedbackTrace>(`/api/feedback-traces/${state.traceId}`), before = state.traceBefore as Record<string, unknown>;
|
|
state.originalTraceUnchanged = ["id", "targetId", "status", "destination", "exportId"].every(key => after[key as keyof FeedbackTrace] === before[key]);
|
|
const sharingAfter = await ctx.input.api.get<{ feedbackDataSharingPreference?: string }>("/api/instance/settings/general");
|
|
state.remoteSharingUnchanged = sharingAfter.feedbackDataSharingPreference === state.sharingPreferenceBefore && before.status === "local_only" && after.status === "local_only" && after.destination == null && after.exportId == null;
|
|
state.bundlePublished = false; state.localOutputDirectory = directory;
|
|
const receipt = { localOutputDirectory: directory, zipSha256: state.zipSha256, zipBytes: state.zipBytes, companyId: company, issueId: task.id,
|
|
feedbackTraceId: state.traceId, originalRunId: state.workerRunId, originalRunStatus: "cancelled", originalAgentId: (state.worker as Entity).id,
|
|
exportedContents: ["scoped feedback vote", "trace metadata", "original cancelled run", "retained run events"], exactLocalBytesVerified: state.fullTraceBytesMatch,
|
|
remoteSharingEnabled: false, publicAttachmentCreated: false, outcomeKind: "operator interrupted unfinished work; not a model failure or failed-native run" };
|
|
await ctx.input.evidence("chat-catalog-local-feedback-export-receipt.json", receipt);
|
|
const message = `The human operator exported this local review bundle and retained the original interrupted work evidence. These are the observed facts; please report its location and contents without publishing it:\n\n${JSON.stringify(receipt)}`;
|
|
await sendChatMessage(ctx.input.page, message, chatCatalogNativePhaseRemainingBudget(ctx)); state.exportReport = await companionPhaseForCancelled(ctx, prior.map(r => r.id), message);
|
|
}, grade: gradeFeedbackBundle, async verifyResult(ctx) { await expect(ctx.input.page.getByTestId("task-chat-composer-input")).toBeVisible(); },
|
|
});
|
|
/** The chat composer saves Markdown, escaping literal punctuation such as the
|
|
* brackets of a JSON array ("\\["). Compare the sent text with that escaping removed. */
|
|
export function unescapeComposerMarkdown(body: string) { return body.replace(/\\([\\`*_{}\[\]()#+\-.!|>~])/g, "$1"); }
|
|
// companionPhase's default all-succeeded assertion intentionally does not apply to this retained cancelled fixture.
|
|
async function companionPhaseForCancelled(ctx: Parameters<typeof companionPhase>[0], prior: string[], message: string) {
|
|
const result = await pollUntil({ label: "Native reply to local interrupted-work export receipt", deadlineAt: chatCatalogNativePhaseDeadline(ctx, Date.now() + 150000),
|
|
load: async () => { await ctx.refreshIssue(); return { issue: ctx.issue(), runs: await ctx.allRuns() }; },
|
|
accept: o => chatCatalogNativePhaseIsTimely(ctx) && (o.issue?.conversationState === "waiting" && o.runs.length === prior.length + 1 && o.runs.filter(r => r.status === "succeeded").length === 2 && o.runs.filter(r => r.status === "cancelled").length === 1) });
|
|
const run = result.runs.find(r => !prior.includes(r.id) && r.agentId === ctx.input.fixtures.agent.id)!;
|
|
const comments = await ctx.input.api.get<Entity[]>(`/api/issues/${ctx.issue()!.id}/comments?order=asc`), human = comments.filter(c => !c.authorAgentId && unescapeComposerMarkdown(c.body ?? "") === message);
|
|
const replies = comments.filter(c => c.authorAgentId === ctx.input.fixtures.agent.id && c.createdByRunId === run.id);
|
|
const profile = await ctx.input.api.get<{ runnerProfileJson?: { nativeExecutionInput?: { task?: { prompt?: string } } } }>(`/api/heartbeat-runs/${run.id}`);
|
|
return { humanCommentId: human[0]?.id, runId: run.id, reply: replies.map(c => c.body).join("\n"),
|
|
attributed: human.length === 1 && replies.length > 0 && profile.runnerProfileJson?.nativeExecutionInput?.task?.prompt?.includes(human[0]!.id) === true
|
|
&& replies.every(c => Date.parse(c.createdAt!) >= Date.parse(human[0]!.createdAt!)) };
|
|
}
|
|
feedback.oracleVersion = "paperclip.chat-catalog.feedback-bundle.v2"; feedback.expectedRunCount = 3; feedback.expectedHumanMessageCount = 2; feedback.attemptTimeoutMs = 900000;
|
|
feedback.runPolicy = { kind: "terminal-counts", terminalStatusCounts: { succeeded: 2, cancelled: 1 } };
|
|
feedback.cleanup = async (ctx, state) => { try { const pending = pendingFixtures.get(ctx);
|
|
if (pending) { const runs = await ctx.allRuns(); for (const run of runs.filter(r => r.agentId === pending.gate.worker.id && ["queued", "running"].includes(r.status))) await ctx.input.api.post(`/api/heartbeat-runs/${run.id}/cancel`, {}); await releaseRuntimeWorker({ gate: pending.gate.gate }); pendingFixtures.delete(ctx); }
|
|
await releaseRuntimeWorker(state); } finally { try { await cleanupIdentityProbes(ctx); } finally { await cleanupOwnedCli(ctx); } } };
|
|
feedback.sourceFiles = [...feedback.sourceFiles, "chat-catalog-domains/privileged-cli-support.ts", "chat-catalog-domains/identity-support.ts", "chat-catalog-domains/runtime-busy-worker.ts", "../../cli/src/index.ts", "../../cli/src/commands/client/feedback.ts", "../../server/src/services/feedback.ts"];
|
|
export const privilegedFeedbackCases = [feedback];
|