mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-13 07:12:00 +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>
94 lines
9.2 KiB
TypeScript
94 lines
9.2 KiB
TypeScript
/** Explicit infrastructure-only operation. Importing or typechecking never
|
|
* starts Docker, a database, Paperclip, a scenario, a model or a native run. */
|
|
import { spawn } from "node:child_process";
|
|
import { createWriteStream } from "node:fs";
|
|
import { mkdir, readFile, rm, writeFile } from "node:fs/promises";
|
|
import { randomBytes } from "node:crypto";
|
|
import path from "node:path";
|
|
import { fileURLToPath, pathToFileURL } from "node:url";
|
|
import { createRunnerE2ETemporaryRoot } from "./server-config.js";
|
|
import { reserveRunnerE2EServerPort } from "./ports.js";
|
|
import { DATABASE_LANE_ENV, DOCKER_DATABASE_ENV, dockerDatabaseLaunchState, publicDockerDatabaseIdentity, runDocker, startOwnedDockerDatabase } from "./docker-postgres.js";
|
|
import { createProcessTreeOwner, stopOwnedProcessTree } from "./process-tree-owner.js";
|
|
import { runnerE2ETypeScriptProcessArgs } from "./web-server-command.js";
|
|
import { assertRunnerDatabaseIsolation } from "./instance-isolation.js";
|
|
import { resolveRunnerE2ESource } from "./source.js";
|
|
import { findSecretLeakInDirectory, sanitizeJson } from "./redaction.js";
|
|
|
|
export async function runOwnedDockerInfrastructureHealth() {
|
|
const source = resolveRunnerE2ESource();
|
|
if (!source.sha || !/^[a-f0-9]{40}$/.test(source.sha)) throw new Error("Infrastructure health requires the explicit immutable source SHA");
|
|
const repositoryRoot = fileURLToPath(new URL("../../", import.meta.url));
|
|
const root = await createRunnerE2ETemporaryRoot(), privateDir = path.join(root, "infrastructure-health-private");
|
|
await mkdir(privateDir, { mode: 0o700 });
|
|
const log = createWriteStream(path.join(privateDir, "server.log"), { mode: 0o600 });
|
|
const secrets: string[] = [], startedAt = new Date().toISOString();
|
|
let database: Awaited<ReturnType<typeof startOwnedDockerDatabase>> | undefined;
|
|
let child: ReturnType<typeof spawn> | undefined, owner: ReturnType<typeof createProcessTreeOwner> | undefined, observation: ReturnType<typeof setInterval> | undefined;
|
|
let persistedStateScanPassed = false;
|
|
let phase = "owned-postgres-startup", failure: unknown; const cleanupFailures: string[] = [];
|
|
let health: { status: string; deploymentMode?: string } | undefined, schema: Record<string, number> | undefined;
|
|
try {
|
|
database = await startOwnedDockerDatabase({ temporaryRoot: root, privateSecrets: secrets });
|
|
phase = "normal-source-onboarding-migrations";
|
|
const port = await reserveRunnerE2EServerPort(), home = path.join(root, "paperclip-home"), instance = `runner-e2e-docker-health-${randomBytes(6).toString("hex")}`;
|
|
const configPath = path.join(home, "instances", instance, "config.json");
|
|
const env: NodeJS.ProcessEnv = Object.fromEntries(["PATH", "HOME", "TMPDIR", "DOCKER_CONFIG", "DOCKER_CONTEXT", "DOCKER_HOST"].filter(key => process.env[key]).map(key => [key, process.env[key]]));
|
|
Object.assign(env, { NODE_ENV: "test", DO_NOT_TRACK: "1", PAPERCLIP_HOME: home, PAPERCLIP_CONFIG: configPath, PAPERCLIP_INSTANCE_ID: instance,
|
|
PAPERCLIP_RUNNER_E2E_SOURCE_SHA: source.sha, PAPERCLIP_RUNNER_E2E_TEMP_ROOT: root, PAPERCLIP_RUNNER_E2E_PORT: String(port), PAPERCLIP_RUNNER_E2E_EXECUTION_IDS: "[]", PAPERCLIP_RUNNER_E2E_SERVER_LOG: path.join(privateDir, "supervisor.log"),
|
|
[DATABASE_LANE_ENV]: "owned-docker-postgres", [DOCKER_DATABASE_ENV]: database.filename });
|
|
for (const name of ["PAPERCLIP_AGENT_JWT_SECRET", "PAPERCLIP_DECISION_SIGNING_SECRET", "PAPERCLIP_TOOL_ACTION_SIGNING_SECRET", "BETTER_AUTH_SECRET"]) { const value = randomBytes(48).toString("hex"); secrets.push(value); env[name] = value; }
|
|
child = spawn(process.execPath, runnerE2ETypeScriptProcessArgs(repositoryRoot, path.join(repositoryRoot, "tests/runner-e2e/server.ts")), { cwd: repositoryRoot, env, detached: true, stdio: ["ignore", "pipe", "pipe"] });
|
|
let spawnFailure = false; child.on("error", () => { spawnFailure = true; });
|
|
owner = createProcessTreeOwner(child); await owner.observe(); observation = setInterval(() => { void owner!.observe().catch(() => {}); }, 100);
|
|
child.stdout?.pipe(log, { end: false }); child.stderr?.pipe(log, { end: false });
|
|
const deadline = Date.now() + 180000, origin = `http://127.0.0.1:${port}`;
|
|
while (Date.now() < deadline) {
|
|
if (spawnFailure || child.exitCode !== null || child.signalCode !== null) throw new Error("Owned source supervisor exited before API readiness; private logs retained");
|
|
try { const response = await fetch(`${origin}/api/health`, { signal: AbortSignal.timeout(2000) }); const value = await response.json() as typeof health; if (response.ok && value?.status === "ok") { health = value; break; } } catch { /* Keep the genuine readiness deadline; no fabricated response. */ }
|
|
await new Promise(resolve => setTimeout(resolve, 250));
|
|
}
|
|
if (!health) throw new Error("Owned source onboarding/API readiness deadline expired");
|
|
phase = "actual-api-schema-and-isolation";
|
|
const companies = await fetch(`${origin}/api/companies`, { signal: AbortSignal.timeout(5000) }); const companyRows = await companies.json();
|
|
const ui = await fetch(origin, { signal: AbortSignal.timeout(5000) });
|
|
if (!companies.ok || !Array.isArray(companyRows) || companyRows.length !== 0 || !ui.ok) throw new Error("Fresh source instance did not expose healthy empty-company API/UI");
|
|
await assertRunnerDatabaseIsolation(configPath, root, database.filename);
|
|
phase = "same-final-persisted-state-secret-scan";
|
|
const leak = await findSecretLeakInDirectory(home, secrets, { includeShapes: false, privateDockerDatabaseConfig: { temporaryRoot: root, configPath, receiptPath: database.filename } });
|
|
if (leak) throw new Error("Owned infrastructure persisted-state scan detected a secret outside intended custody");
|
|
persistedStateScanPassed = true;
|
|
await database.verify();
|
|
// Only the exact already-owned CID can receive this read-only SQL command.
|
|
const r = database.receipt;
|
|
schema = JSON.parse(await runDocker(["--context", r.engine.context, "exec", r.containerId!, "psql", "-X", "-U", r.username, "-d", r.database, "-At", "-c",
|
|
"SELECT json_build_object('migrations',(SELECT count(*) FROM drizzle.__drizzle_migrations),'companies',(SELECT count(*) FROM companies),'runs',(SELECT count(*) FROM heartbeat_runs),'agents',(SELECT count(*) FROM agents))"]));
|
|
if (!schema || !Number.isSafeInteger(schema.migrations) || schema.migrations <= 0 || schema.companies !== 0 || schema.agents !== 0 || schema.runs !== 0) throw new Error("Fresh normal migrations/zero-work SQL assertions failed");
|
|
phase = "verified-zero-work-infrastructure";
|
|
} catch (error) {
|
|
failure = error; await writeFile(path.join(privateDir, "primary-error.txt"), error instanceof Error ? error.stack ?? error.message : "Infrastructure health failed", { mode: 0o600 });
|
|
} finally {
|
|
if (observation) clearInterval(observation);
|
|
if (child && owner) try { await stopOwnedProcessTree(child, owner); } catch { cleanupFailures.push("Owned source supervisor/descendant cleanup unconfirmed"); }
|
|
// Never stop a database beneath a source process whose exit is unconfirmed.
|
|
if (database && !cleanupFailures.length) try { await database.close(); } catch { cleanupFailures.push("Owned Docker container/volume cleanup unconfirmed"); }
|
|
await new Promise<void>(resolve => log.end(resolve));
|
|
}
|
|
const startupState = await dockerDatabaseLaunchState("owned-docker-postgres", root);
|
|
if (startupState.cleanupUnconfirmed && !cleanupFailures.length) cleanupFailures.push("Owned Docker resource absence is unconfirmed");
|
|
const receipt = { schemaVersion: 1, kind: "owned-docker-postgres-infrastructure-only", source, productionBaseline: "8373d7867963674ff960d2345d0d28086f799bc7", startedAt, finishedAt: new Date().toISOString(),
|
|
status: failure || cleanupFailures.length ? "failed" : "passed", phase, primaryFailure: failure ? "Actual infrastructure operation failed; primary private diagnostics retained" : null, cleanupFailures,
|
|
database: database ? publicDockerDatabaseIdentity(database.receipt) : startupState.identity,
|
|
checks: { sameFinalPersistedStateSecretScan: persistedStateScanPassed, actualApiHealth: !!health, actualNormalMigrations: !!schema?.migrations, zeroCompanies: schema?.companies === 0, zeroNativeRuns: schema?.runs === 0, zeroAgents: schema?.agents === 0, zeroScenarios: true, zeroModels: true },
|
|
canonicalAttempts: 0, privateRoot: failure || cleanupFailures.length ? root : undefined };
|
|
const destination = path.join(repositoryRoot, "tests/runner-e2e/results/chat-catalog-campaign"); await mkdir(destination, { recursive: true });
|
|
const filename = path.join(destination, `docker-infrastructure-health-${startedAt.replaceAll(/[:.]/g, "-")}.json`);
|
|
await writeFile(filename, `${JSON.stringify(sanitizeJson(receipt, secrets), null, 2)}\n`, { mode: 0o600 });
|
|
if (!failure && !cleanupFailures.length) await rm(root, { recursive: true, force: true });
|
|
return { status: receipt.status, receipt: filename };
|
|
}
|
|
if (process.argv[1] && import.meta.url === pathToFileURL(path.resolve(process.argv[1])).href) {
|
|
if (process.argv.slice(2).join() !== "--run-owned-infrastructure-health") throw new Error("Requires explicit --run-owned-infrastructure-health; never a canonical scenario");
|
|
const result = await runOwnedDockerInfrastructureHealth(); process.stdout.write(`${JSON.stringify(result)}\n`); if (result.status !== "passed") process.exitCode = 1;
|
|
}
|