Files
PaperClipAI/server/src/__tests__/server-startup-feedback-export.test.ts
T
cad26c6bfb fix(tool-gateway): bound MCP discovery memory and concurrency (#14864)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - Agents discover governed tools through the MCP gateway.
> - A listing repeated policy and full run-row reads for each catalog
tool.
> - Parallel listings multiplied those allocations during run startup.
> - One 900-tool baseline listing used 11,489 queries and about 1.7 GiB
of extra heap in a fixture.
> - This pull request shares reads within a listing and bounds whole
listings across the process.
> - The benefit is lower discovery memory use while execution still
checks current policy.

## Linked Issues or Issue Description

Refs #13115. Its on-demand target change affects the same listing loop.

**What happened?**

MCP discovery repeated roughly 13 reads per tool. Full run snapshots and
repeated connection configurations caused large allocations. Per-listing
bounds alone did not limit concurrent listings across gateways.

**Expected behavior**

Discovery reads shared inputs once per listing. The process bounds
active and queued listings. Catalog payload size and policy evaluation
still grow with the catalog. Tool execution checks current access rules.

**Steps to reproduce**

Create a company with a remote MCP connection, 900 catalog tools, large
schemas, and a large run snapshot. Send concurrent tools/list requests
using a run-bound gateway token. Run the committed benchmark for a
deterministic reproduction.

**Paperclip version or commit**

Baseline: f2e0f19630. Measured on Node
26.4.0 on macOS.

## What Changed

- Keep Michael Nguyen's per-listing policy cache, scalar policy reads,
16-decision bound, and catalog/connection split under repeatable read.
- Admit two whole listings per process and queue at most 32. Return 503
with tool_discovery_busy when full.
- Propagate disconnects to discovery. Stop scheduling new reads and
drain started reads before freeing the slot.
- Project context IDs in gateway authentication and GitHub runtime
discovery. Avoid loading task descriptions and results.
- Keep discovery audit counts and a SHA-256 digest instead of the full
name list. Retain access and activity audit records.
- Sweep expired named gateway tokens at startup and on the existing
scheduler. Delete at most 500 per pass. Add an idempotent expiry-index
migration. Resolve audit token references atomically so cleanup cannot
break admitted requests.
- Return GET 405 with Allow: POST on stateless MCP gateway and
runtime-tools endpoints. Keep runtime-tools authentication.
- Add red-green regressions, HTTP concurrency and policy-revocation
coverage, and browser approval assertions.
- Commit the benchmark harness, raw results, and resource-bound
documentation.

## Verification

- Baseline listing regressions failed with 722 queries for 50 tools and
6,422 for 500. The connection-row duplication regression also failed.
- Follow-up regressions failed before the fixes for full snapshot reads,
abandoned listings, full name-list audits, and expired tokens.
- Listing and scheduler tests: 14 pass. Existing gateway/policy suites
passed after preserving the cleanup return contract.
- Two HTTP journeys pass: initialize, GET/SSE rejection, 16 concurrent
500-tool listings, provider call, policy revocation, and denied retry.
Token cleanup during provider dispatch also completes successfully and
blocks subsequent requests.
- pnpm test:e2e tests/e2e/mcp-user-stories.spec.ts --grep
'@mcp-runnable': 8 pass. The approval journey clicks Allow once in the
browser. Review screenshots wait for loaded content.
- pnpm -r typecheck: passes. pnpm build: passes.
- The general server group completed with 14,755 passes and 18 failures.
The 17 startup mock failures were fixed; all 21 startup tests pass on
rerun. The one Discord timing failure passed on the unchanged baseline
and on rerun (74 tests). UI and CLI groups pass 7,632 tests. Shared and
skill groups pass 853 tests. The remaining database and adapter groups
pass 3,010 tests with one worker after a macOS shared-memory limit
interrupted a parallel run. All 149 serialized route files pass (2,762
tests).
- At 900 tools, one listing falls from 11,489 to 36 queries and from
about 1.7 GiB to 37 MiB of extra heap. Sixteen concurrent listings used
169–187 MiB of extra heap. Four connections used 39 queries per listing.
- Latest-head verification: 54 successful checks and two expected skips
on 5c0793090c. Fresh Greptile review: 5/5
with no unresolved threads.
- Reproduce with server/scripts/benchmark-tool-gateway-listing.ts. See
doc/mcp-discovery-performance.md and
doc/benchmarks/2026-10-01-mcp-discovery.json.

## Risks

- The process-wide FIFO queue can increase discovery latency. Excess
callers must retry 503 responses.
- Cancellation applies to discovery. Started database reads finish
before their slot is released.
- Audit consumers must use visibleToolCount and visibleToolsHash instead
of visibleTools.
- The expiry index can briefly lock the token table during migration.
Sweeps preserve unexpired and non-expiring tokens.
- Measurements use isolated fixtures and deterministic providers. They
do not establish a production heap limit or affected installation count.
Rate-limit reads remain uncached.

## Model Used

- Original listing optimization: Anthropic Claude Opus 5.5
(claude-opus-5-5), Claude Code, extended thinking and tool use, as
recorded by the original author.
- Follow-up fixes and verification: OpenAI Codex, GPT-6, with shell
execution, file edits, database fixtures, and browser tests. The exact
serving model ID and context window are not exposed in this session.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with Fixes / Closes /
Refs OR (b) described the issue in-PR following the relevant issue
template
- [x] I have not referenced internal/instance-local Paperclip issues or
links
- [x] My branch name describes the change 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
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Dotta <bippadotta@protonmail.com>
Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-10-02 07:45:27 -05:00

895 lines
34 KiB
TypeScript

import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
import { chmodSync, mkdirSync, mkdtempSync, readFileSync, rmSync, statSync, symlinkSync, writeFileSync } from "node:fs";
import { tmpdir } from "node:os";
import path from "node:path";
const ORIGINAL_PAPERCLIP_API_URL = process.env.PAPERCLIP_API_URL;
const ORIGINAL_PAPERCLIP_RUNTIME_API_URL = process.env.PAPERCLIP_RUNTIME_API_URL;
const ORIGINAL_PAPERCLIP_RUNTIME_API_CANDIDATES_JSON = process.env.PAPERCLIP_RUNTIME_API_CANDIDATES_JSON;
const ORIGINAL_PAPERCLIP_LISTEN_HOST = process.env.PAPERCLIP_LISTEN_HOST;
const ORIGINAL_PAPERCLIP_LISTEN_PORT = process.env.PAPERCLIP_LISTEN_PORT;
const {
completionSweepMock,
createAppMock,
createBetterAuthInstanceMock,
createDbMock,
detectPortMock,
deriveAuthTrustedOriginsMock,
environmentCustomImagesServiceMock,
environmentCustomImagesServiceFactoryMock,
executionWorkspaceServiceFactoryMock,
executionWorkspaceServiceMock,
externalObjectsServiceMock,
externalObjectsServiceFactoryMock,
feedbackExportServiceMock,
feedbackServiceFactoryMock,
fakeServer,
heartbeatServiceFactoryMock,
heartbeatServiceMock,
issueThreadInteractionServiceFactoryMock,
issueThreadInteractionServiceMock,
loadConfigMock,
resolveHeartbeatSchedulingSuppressionMock,
routineServiceFactoryMock,
routineServiceMock,
} = vi.hoisted(() => {
const completionSweepMock = vi.fn(async () => undefined);
const createAppMock = vi.fn(async () => Object.assign((_: unknown, __: unknown) => {}, {
locals: {
toolGateway: {
sweepActionReviews: vi.fn(async () => ({ scanned: 0 })),
cleanupExpiredSessions: vi.fn(async () => ({ deletedCount: 0 })),
},
toolActionDeliveries: { sweepPending: vi.fn(async () => ({ scanned: 0, delivered: 0 })) },
},
}) as never);
const createBetterAuthInstanceMock = vi.fn(() => ({}));
const createDbMock = vi.fn(() => ({
select: vi.fn(() => ({
from: vi.fn(() => ({ where: vi.fn(async () => []) })),
})),
}) as never);
const detectPortMock = vi.fn(async ({ port }: { port: number; hostname: string }) => port);
const deriveAuthTrustedOriginsMock = vi.fn(() => []);
const resolveHeartbeatSchedulingSuppressionMock = vi.fn(() => ({
suppressed: false,
reason: null,
}));
const heartbeatServiceMock = {
resolveSchedulingSuppression: resolveHeartbeatSchedulingSuppressionMock,
recoverNativeRunsAfterRestart: vi.fn(async () => ({
restartKind: "hard",
dispositions: [],
claims: [],
awaitingEvidenceRunIds: [],
blockedRunIds: [],
})),
reconcileHotRestartAdoption: vi.fn(async () => ({ mode: "none" })),
reapOrphanedRuns: vi.fn(async () => ({ reaped: 0, runIds: [] })),
promoteDueScheduledRetries: vi.fn(async () => ({ promoted: 0, runIds: [] })),
resumeQueuedRuns: vi.fn(async () => undefined),
recoverPendingSessionGoalActions: vi.fn(async () => ({ scanned: 0, enqueued: 0, alreadyQueued: 0, invalid: 0 })),
recoverActiveSessionGoals: vi.fn(async () => ({ scanned: 0, enqueued: 0 })),
reconcileStrandedAssignedIssues: vi.fn(async () => ({
assignmentDispatched: 0,
dispatchRequeued: 0,
continuationRequeued: 0,
successfulRunHandoffEscalated: 0,
escalated: 0,
skipped: 0,
issueIds: [],
})),
reconcileResolvedDependencyWakes: vi.fn(async () => ({ healed: 0 })),
reconcileTaskWatchdogs: vi.fn(async () => ({ triggered: 0 })),
scanSilentActiveRuns: vi.fn(async () => ({ created: 0, escalated: 0 })),
sweepStaleIssueLocks: vi.fn(async () => ({ cleared: 0 })),
sweepPendingCleanupLeases: vi.fn(async () => ({ swept: 0, destroyed: 0, capped: 0 })),
sweepExpiredRuntimeStatuses: vi.fn(() => 0),
tickTimers: vi.fn(async () => ({ checked: 0, enqueued: 0, skipped: 0 })),
};
const heartbeatServiceFactoryMock = vi.fn(() => heartbeatServiceMock);
const issueThreadInteractionServiceMock = {
sweepSupersededPendingRequestConfirmations: vi.fn(async () => ({ expired: 0 })),
sweepMergedPullRequestConfirmations: vi.fn(async () => ({
checked: 0,
candidates: 0,
accepted: 0,
woken: 0,
})),
};
const issueThreadInteractionServiceFactoryMock = vi.fn(() => issueThreadInteractionServiceMock);
const environmentCustomImagesServiceMock = {
cleanupExpiredSetupSessions: vi.fn(async () => ({ scanned: 0, timedOut: 0, failed: 0 })),
};
const environmentCustomImagesServiceFactoryMock = vi.fn(() => environmentCustomImagesServiceMock);
const executionWorkspaceServiceMock = {
sweepTerminalWorkspaces: vi.fn(async () => ({
checked: 0,
eligible: 0,
archived: 0,
cleanupFailed: 0,
skippedActiveRun: 0,
skippedNonTerminalTree: 0,
skippedUndelivered: 0,
skippedRace: 0,
})),
};
const executionWorkspaceServiceFactoryMock = vi.fn(() => executionWorkspaceServiceMock);
const externalObjectsServiceMock = {
refreshDueObjectsForActiveCompanies: vi.fn(async () => ({ companies: 0, checked: 0, refreshed: 0 })),
};
const externalObjectsServiceFactoryMock = vi.fn(() => externalObjectsServiceMock);
const routineServiceMock = {
tickScheduledTriggers: vi.fn(async () => ({ triggered: 0 })),
};
const routineServiceFactoryMock = vi.fn(() => routineServiceMock);
const feedbackExportServiceMock = {
flushPendingFeedbackTraces: vi.fn(async () => ({ attempted: 0, sent: 0, failed: 0 })),
};
const feedbackServiceFactoryMock = vi.fn(() => feedbackExportServiceMock);
const fakeServer = {
on: vi.fn().mockReturnThis(),
once: vi.fn().mockReturnThis(),
off: vi.fn().mockReturnThis(),
listen: vi.fn((_port: number, _host: string, callback?: () => void) => {
callback?.();
return fakeServer;
}),
close: vi.fn((callback?: (error?: Error) => void) => callback?.()),
};
const loadConfigMock = vi.fn();
return {
completionSweepMock,
createAppMock,
createBetterAuthInstanceMock,
createDbMock,
detectPortMock,
deriveAuthTrustedOriginsMock,
environmentCustomImagesServiceMock,
environmentCustomImagesServiceFactoryMock,
executionWorkspaceServiceFactoryMock,
executionWorkspaceServiceMock,
externalObjectsServiceMock,
externalObjectsServiceFactoryMock,
feedbackExportServiceMock,
feedbackServiceFactoryMock,
fakeServer,
heartbeatServiceFactoryMock,
heartbeatServiceMock,
issueThreadInteractionServiceFactoryMock,
issueThreadInteractionServiceMock,
loadConfigMock,
resolveHeartbeatSchedulingSuppressionMock,
routineServiceFactoryMock,
routineServiceMock,
};
});
function buildTestConfig(overrides: Record<string, unknown> = {}) {
return {
deploymentMode: "authenticated",
deploymentExposure: "private",
bind: "loopback",
customBindHost: undefined,
host: "127.0.0.1",
port: 3210,
allowedHostnames: [],
authBaseUrlMode: "auto",
authPublicBaseUrl: undefined,
authDisableSignUp: false,
databaseMode: "postgres",
databaseUrl: "postgres://paperclip:paperclip@127.0.0.1:5432/paperclip",
embeddedPostgresDataDir: "/tmp/paperclip-test-db",
embeddedPostgresPort: 54329,
databaseBackupEnabled: false,
databaseBackupIntervalMinutes: 60,
databaseBackupRetentionDays: 30,
databaseBackupDir: "/tmp/paperclip-test-backups",
serveUi: false,
uiDevMiddleware: false,
secretsProvider: "local_encrypted",
secretsStrictMode: false,
secretsMasterKeyFilePath: "/tmp/paperclip-master.key",
storageProvider: "local_disk",
storageLocalDiskBaseDir: "/tmp/paperclip-storage",
storageS3Bucket: "paperclip-test",
storageS3Region: "us-east-1",
storageS3Endpoint: undefined,
storageS3Prefix: "",
storageS3ForcePathStyle: false,
feedbackExportBackendUrl: "https://telemetry.example.com",
feedbackExportBackendToken: "telemetry-token",
heartbeatSchedulerEnabled: false,
heartbeatSchedulerIntervalMs: 30000,
companyDeletionEnabled: false,
...overrides,
};
}
vi.mock("node:http", () => ({
createServer: vi.fn(() => fakeServer),
}));
vi.mock("detect-port", () => ({
default: detectPortMock,
}));
vi.mock("@paperclipai/db", () => ({
createDb: createDbMock,
ensurePostgresDatabase: vi.fn(),
getPostgresDataDirectory: vi.fn(),
inspectMigrations: vi.fn(async () => ({ status: "upToDate" })),
applyPendingMigrations: vi.fn(),
reconcilePendingMigrationHistory: vi.fn(async () => ({ repairedMigrations: [] })),
formatDatabaseBackupResult: vi.fn(() => "ok"),
runDatabaseBackup: vi.fn(),
authUsers: {},
companies: {},
companyMemberships: {},
instanceUserRoles: {},
}));
vi.mock("../app.js", () => ({
createApp: createAppMock,
}));
vi.mock("../services/native-runtime/native-session-executor.js", () => ({
verifyStoppedNativeSessionForReplacement: vi.fn(async () => null),
}));
// This suite verifies server startup scheduling; replacement correctness is
// exercised by the dedicated DB-backed recovery suites.
vi.mock("../services/native-runtime/native-safe-replacement.js", () => ({
reconcileSafeNativeReplacements: vi.fn(async () => ({ scanned: 0, scheduled: 0 })),
}));
vi.mock("../config.js", () => ({
loadConfig: loadConfigMock,
}));
vi.mock("../middleware/logger.js", () => ({
logger: {
child: vi.fn(function child() {
return this;
}),
info: vi.fn(),
warn: vi.fn(),
error: vi.fn(),
},
}));
vi.mock("../realtime/live-events-ws.js", () => ({
setupLiveEventsWebSocketServer: vi.fn(),
}));
vi.mock("../services/index.js", () => ({
backfillLegacyToolOAuthTokens: vi.fn(async () => ({
scannedConnections: 0,
migratedConnections: 0,
sanitizedConnections: 0,
createdSecrets: 0,
rotatedSecrets: 0,
accessTokensBackfilled: 0,
refreshTokensBackfilled: 0,
})),
backfillPrincipalAccessCompatibility: vi.fn(async () => ({
agentMembershipsInserted: 0,
humanGrantsInserted: 0,
})),
attentionService: vi.fn(() => ({
list: vi.fn(async () => ({ items: [], nextCursor: null })),
})),
decisionService: vi.fn(() => ({
sweepExpired: vi.fn(async () => ({ expired: 0 })),
})),
decisionRetentionService: vi.fn(() => ({
autoArchive: vi.fn(async () => 0),
deliverNotifications: vi.fn(async () => ({ notifiedAgents: 0, delivered: 0 })),
})),
feedbackService: feedbackServiceFactoryMock,
bootstrapExecutionPolicyFromEnv: vi.fn(async () => null),
applyManagedEnvironments: vi.fn(async () => null),
environmentCustomImageService: environmentCustomImagesServiceFactoryMock,
executionWorkspaceService: executionWorkspaceServiceFactoryMock,
externalObjectService: externalObjectsServiceFactoryMock,
heartbeatService: heartbeatServiceFactoryMock,
githubConnectionEventService: vi.fn(() => ({
pollOnce: vi.fn(async () => ({
leased: 0,
processed: 0,
duplicate: 0,
ignored: 0,
failed: 0,
})),
})),
issueThreadInteractionService: issueThreadInteractionServiceFactoryMock,
issueService: vi.fn(() => ({ update: vi.fn(async () => null) })),
instanceSettingsService: vi.fn(() => ({
getExperimental: vi.fn(async () => ({
enableExternalObjects: true,
enableStatusCards: false,
})),
getGeneral: vi.fn(async () => ({
backupRetention: {
dailyDays: 7,
weeklyWeeks: 4,
monthlyMonths: 1,
},
})),
})),
reconcileCodexLocalManagedHomesOnStartup: vi.fn(async () => ({
scanned: 0,
seeded: 0,
alreadySeeded: 0,
externalOverride: 0,
noManagedHome: 0,
sourceAuthMissing: 0,
failed: 0,
seededAgentIds: [],
})),
reconcileBuiltInAgentsOnStartup: vi.fn(async () => ({
scanned: 0,
reconciled: 0,
unknown: 0,
duplicates: 0,
})),
reconcilePersistedRuntimeServicesOnStartup: vi.fn(async () => ({ reconciled: 0 })),
resolveHeartbeatSchedulingSuppression: resolveHeartbeatSchedulingSuppressionMock,
routineService: routineServiceFactoryMock,
statusCardService: vi.fn(() => ({})),
toolAccessService: vi.fn(() => ({
sweepConnectionHealth: vi.fn(async () => ({
checked: 0,
healthy: 0,
needsAttention: 0,
failed: 0,
})),
sweepGitHubConnectionContinuity: vi.fn(async () => ({
checked: 0,
due: 0,
refreshed: 0,
failed: 0,
})),
})),
}));
vi.mock("../services/chat-completion-delivery.js", () => ({ chatCompletionDeliveryService: () => ({ sweepPending: completionSweepMock }) }));
vi.mock("../services/connection-intent-delivery.js", () => ({
connectionIntentDeliveryService: vi.fn(() => ({
sweepPending: vi.fn(async () => ({ scanned: 0, failed: 0 })),
})),
}));
vi.mock("../services/question-response-delivery.js", () => ({
questionResponseDeliveryService: vi.fn(() => ({
sweepPending: vi.fn(async () => ({
scanned: 0,
steered: 0,
coalesced: 0,
wakeFallback: 0,
failed: 0,
})),
})),
}));
vi.mock("../services/native-runtime/native-question-bridge.js", () => ({
deliverNativeQuestionResponse: vi.fn(async () => "not_native"),
nativeQuestionCancellationIdentity: vi.fn(() => null),
nativeQuestionRunToCancel: vi.fn(async () => null),
validateNativeQuestionResponseInput: vi.fn(),
}));
vi.mock("../services/secret-proposals.js", () => ({
createSecretProposalsService: vi.fn(() => ({
sweepExpired: vi.fn(async () => 0),
})),
}));
vi.mock("../storage/index.js", () => ({
createStorageServiceFromConfig: vi.fn(() => ({ id: "storage-service" })),
}));
vi.mock("../services/feedback-share-client.js", () => ({
createFeedbackTraceShareClientFromConfig: vi.fn(() => ({ id: "feedback-share-client" })),
}));
vi.mock("../services/plugin-worker-manager.js", () => ({
createPluginWorkerManager: vi.fn(() => ({ id: "plugin-worker-manager" })),
}));
vi.mock("../startup-banner.js", () => ({
printStartupBanner: vi.fn(),
}));
vi.mock("../board-claim.js", () => ({
getBoardClaimWarningUrl: vi.fn(() => null),
initializeBoardClaimChallenge: vi.fn(async () => undefined),
}));
vi.mock("../auth/better-auth.js", () => ({
createBetterAuthHandler: vi.fn(() => undefined),
createBetterAuthInstance: createBetterAuthInstanceMock,
deriveAuthTrustedOrigins: deriveAuthTrustedOriginsMock,
resolveBetterAuthSession: vi.fn(async () => null),
resolveBetterAuthSessionFromHeaders: vi.fn(async () => null),
}));
import { startServer } from "../index.ts";
import { reconcileSafeNativeReplacements } from "../services/native-runtime/native-safe-replacement.js";
import { EXECUTION_RECONCILIATION_INTERVAL_MS } from "../services/execution-control-deadline.js";
describe("startServer feedback export wiring", () => {
beforeEach(() => {
vi.clearAllMocks();
process.env.PAPERCLIP_DECISION_SIGNING_SECRET = "fedcba9876543210fedcba9876543210";
process.env.PAPERCLIP_AGENT_JWT_SECRET = "0123456789abcdef0123456789abcdef";
loadConfigMock.mockReturnValue(buildTestConfig());
resolveHeartbeatSchedulingSuppressionMock.mockReturnValue({
suppressed: false,
reason: null,
});
createBetterAuthInstanceMock.mockReturnValue({});
deriveAuthTrustedOriginsMock.mockReturnValue([]);
process.env.BETTER_AUTH_SECRET = "test-secret";
});
it("starts without PAPERCLIP_DECISION_SIGNING_SECRET by generating a persisted key", async () => {
const originalHome = process.env.PAPERCLIP_HOME;
const originalInstanceId = process.env.PAPERCLIP_INSTANCE_ID;
const tempHome = mkdtempSync(path.join(tmpdir(), "paperclip-decision-key-"));
process.env.PAPERCLIP_HOME = tempHome;
process.env.PAPERCLIP_INSTANCE_ID = "default";
delete process.env.PAPERCLIP_DECISION_SIGNING_SECRET;
try {
const started = await startServer();
expect(started.server).toBe(fakeServer);
const keyPath = path.join(tempHome, "instances", "default", "secrets", "decision-signing.key");
expect(readFileSync(keyPath, "utf8").trim().length).toBeGreaterThanOrEqual(32);
if (process.platform !== "win32") {
expect(statSync(path.dirname(keyPath)).mode & 0o777).toBe(0o700);
expect(statSync(keyPath).mode & 0o777).toBe(0o600);
}
} finally {
if (originalHome === undefined) delete process.env.PAPERCLIP_HOME;
else process.env.PAPERCLIP_HOME = originalHome;
if (originalInstanceId === undefined) delete process.env.PAPERCLIP_INSTANCE_ID;
else process.env.PAPERCLIP_INSTANCE_ID = originalInstanceId;
rmSync(tempHome, { recursive: true, force: true });
}
});
it("repairs permissive permissions on an existing generated decision signing key", async () => {
const originalHome = process.env.PAPERCLIP_HOME;
const originalInstanceId = process.env.PAPERCLIP_INSTANCE_ID;
const tempHome = mkdtempSync(path.join(tmpdir(), "paperclip-decision-key-mode-"));
const keyPath = path.join(tempHome, "instances", "default", "secrets", "decision-signing.key");
const existingKey = Buffer.alloc(32, 7).toString("base64");
mkdirSync(path.dirname(keyPath), { recursive: true, mode: 0o777 });
chmodSync(path.dirname(keyPath), 0o777);
writeFileSync(keyPath, existingKey, { encoding: "utf8", mode: 0o644 });
chmodSync(keyPath, 0o644);
process.env.PAPERCLIP_HOME = tempHome;
process.env.PAPERCLIP_INSTANCE_ID = "default";
delete process.env.PAPERCLIP_DECISION_SIGNING_SECRET;
try {
const started = await startServer();
expect(started.server).toBe(fakeServer);
expect(readFileSync(keyPath, "utf8")).toBe(existingKey);
if (process.platform !== "win32") {
expect(statSync(path.dirname(keyPath)).mode & 0o777).toBe(0o700);
expect(statSync(keyPath).mode & 0o777).toBe(0o600);
}
} finally {
if (originalHome === undefined) delete process.env.PAPERCLIP_HOME;
else process.env.PAPERCLIP_HOME = originalHome;
if (originalInstanceId === undefined) delete process.env.PAPERCLIP_INSTANCE_ID;
else process.env.PAPERCLIP_INSTANCE_ID = originalInstanceId;
rmSync(tempHome, { recursive: true, force: true });
}
});
it("refuses a symlink planted as the generated decision signing key", async () => {
if (process.platform === "win32") return;
const originalHome = process.env.PAPERCLIP_HOME;
const originalInstanceId = process.env.PAPERCLIP_INSTANCE_ID;
const tempHome = mkdtempSync(path.join(tmpdir(), "paperclip-decision-key-symlink-"));
const keyPath = path.join(tempHome, "instances", "default", "secrets", "decision-signing.key");
const plantedTarget = path.join(tempHome, "planted.key");
const plantedKey = Buffer.alloc(32, 9).toString("base64");
mkdirSync(path.dirname(keyPath), { recursive: true, mode: 0o777 });
chmodSync(path.dirname(keyPath), 0o777);
writeFileSync(plantedTarget, plantedKey, { encoding: "utf8", mode: 0o600 });
symlinkSync(plantedTarget, keyPath);
process.env.PAPERCLIP_HOME = tempHome;
process.env.PAPERCLIP_INSTANCE_ID = "default";
delete process.env.PAPERCLIP_DECISION_SIGNING_SECRET;
try {
await expect(startServer()).rejects.toThrow("must be a regular file");
expect(readFileSync(plantedTarget, "utf8")).toBe(plantedKey);
} finally {
if (originalHome === undefined) delete process.env.PAPERCLIP_HOME;
else process.env.PAPERCLIP_HOME = originalHome;
if (originalInstanceId === undefined) delete process.env.PAPERCLIP_INSTANCE_ID;
else process.env.PAPERCLIP_INSTANCE_ID = originalInstanceId;
rmSync(tempHome, { recursive: true, force: true });
}
});
it("refuses startup when an explicit decision signing secret is too short", async () => {
process.env.PAPERCLIP_DECISION_SIGNING_SECRET = "too-short";
await expect(startServer()).rejects.toThrow("PAPERCLIP_DECISION_SIGNING_SECRET must be at least 32 characters");
expect(loadConfigMock).not.toHaveBeenCalled();
});
it("passes the feedback export service into createApp so pending traces flush in runtime", async () => {
const started = await startServer();
expect(started.server).toBe(fakeServer);
expect(feedbackServiceFactoryMock).toHaveBeenCalledTimes(1);
expect(createAppMock).toHaveBeenCalledTimes(1);
expect(createAppMock.mock.calls[0]?.[1]).toMatchObject({
feedbackExportService: feedbackExportServiceMock,
storageService: { id: "storage-service" },
serverPort: 3210,
});
});
it("keeps startup available when completion delivery recovery fails", async () => {
completionSweepMock.mockRejectedValueOnce(new Error("temporary delivery failure"));
const { startServer } = await import("../index.js");
await expect(startServer()).resolves.toBeDefined();
expect(completionSweepMock).toHaveBeenCalled();
});
it("never invokes the retired review detector at startup or on periodic recovery", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
heartbeatSchedulerEnabled: true,
heartbeatSchedulerIntervalMs: 30000,
}));
const retiredDetector = vi.fn(async () => ({ created: 1, updated: 1, failed: 0 }));
const runtime = Object.assign(heartbeatServiceMock, { reconcileProductivityReviews: retiredDetector });
let intervalCallback: (() => void) | null = null;
const setIntervalSpy = vi.spyOn(globalThis, "setInterval").mockImplementation(((callback: () => void) => {
intervalCallback = callback;
return 1 as unknown as ReturnType<typeof setInterval>;
}) as typeof setInterval);
try {
await startServer();
expect(heartbeatServiceMock.sweepStaleIssueLocks).toHaveBeenCalledTimes(1);
expect(intervalCallback).not.toBeNull();
intervalCallback?.();
await new Promise<void>((resolve) => setImmediate(resolve));
expect(heartbeatServiceMock.sweepStaleIssueLocks).toHaveBeenCalledTimes(2);
expect(retiredDetector).not.toHaveBeenCalled();
} finally {
delete (runtime as Partial<typeof runtime>).reconcileProductivityReviews;
setIntervalSpy.mockRestore();
}
});
it("reconciles native replacements at startup and on the execution-control interval", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({ heartbeatSchedulerEnabled: true }));
let executionControlTick: (() => void) | undefined;
const setIntervalSpy = vi.spyOn(globalThis, "setInterval").mockImplementation(((
callback: () => void,
interval: number,
) => {
if (interval === EXECUTION_RECONCILIATION_INTERVAL_MS) executionControlTick = callback;
return 1 as unknown as ReturnType<typeof setInterval>;
}) as typeof setInterval);
try {
await startServer();
await new Promise<void>((resolve) => setImmediate(resolve));
expect(reconcileSafeNativeReplacements).toHaveBeenCalledExactlyOnceWith(
createDbMock.mock.results[0]?.value,
expect.any(Date),
{ verifyStoppedSession: expect.any(Function) },
);
expect(executionControlTick).toBeDefined();
executionControlTick?.();
await new Promise<void>((resolve) => setImmediate(resolve));
expect(reconcileSafeNativeReplacements).toHaveBeenCalledTimes(2);
} finally {
setIntervalSpy.mockRestore();
}
});
it("keeps routine ticks and setup cleanup active when heartbeat scheduling is suppressed", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
heartbeatSchedulerEnabled: true,
heartbeatSchedulerIntervalMs: 30000,
}));
resolveHeartbeatSchedulingSuppressionMock.mockReturnValue({
suppressed: true,
reason: "worktree_instance",
});
let intervalCallback: (() => void) | null = null;
const setIntervalSpy = vi
.spyOn(globalThis, "setInterval")
.mockImplementation(((callback: () => void) => {
intervalCallback = callback;
return 1 as unknown as ReturnType<typeof setInterval>;
}) as typeof setInterval);
try {
await startServer();
expect(heartbeatServiceMock.reapOrphanedRuns).not.toHaveBeenCalled();
expect(heartbeatServiceMock.tickTimers).not.toHaveBeenCalled();
expect(environmentCustomImagesServiceMock.cleanupExpiredSetupSessions).toHaveBeenCalledTimes(1);
expect(intervalCallback).not.toBeNull();
intervalCallback?.();
await Promise.resolve();
await Promise.resolve();
expect(heartbeatServiceMock.tickTimers).not.toHaveBeenCalled();
expect(externalObjectsServiceMock.refreshDueObjectsForActiveCompanies).toHaveBeenCalledTimes(1);
expect(issueThreadInteractionServiceMock.sweepMergedPullRequestConfirmations).toHaveBeenCalledTimes(1);
expect(executionWorkspaceServiceMock.sweepTerminalWorkspaces).toHaveBeenCalledTimes(1);
expect(routineServiceMock.tickScheduledTriggers).toHaveBeenCalledTimes(1);
expect(environmentCustomImagesServiceMock.cleanupExpiredSetupSessions).toHaveBeenCalledTimes(2);
} finally {
setIntervalSpy.mockRestore();
}
});
it("keeps external object refresh active when heartbeat scheduling is disabled", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
heartbeatSchedulerEnabled: false,
heartbeatSchedulerIntervalMs: 30000,
}));
let intervalCallback: (() => void) | null = null;
const setIntervalSpy = vi
.spyOn(globalThis, "setInterval")
.mockImplementation(((callback: () => void) => {
intervalCallback = callback;
return 1 as unknown as ReturnType<typeof setInterval>;
}) as typeof setInterval);
try {
await startServer();
// The disabled path still creates one heartbeat runtime. This runtime owns
// the orphan-sandbox cleanup sweep, so a leaked provider sandbox is still
// reaped at startup and on the interval.
expect(heartbeatServiceFactoryMock).toHaveBeenCalledTimes(1);
expect(heartbeatServiceMock.sweepPendingCleanupLeases).toHaveBeenCalled();
expect(intervalCallback).not.toBeNull();
intervalCallback?.();
await Promise.resolve();
await Promise.resolve();
expect(externalObjectsServiceMock.refreshDueObjectsForActiveCompanies).toHaveBeenCalledTimes(1);
expect(routineServiceMock.tickScheduledTriggers).not.toHaveBeenCalled();
expect(environmentCustomImagesServiceMock.cleanupExpiredSetupSessions).not.toHaveBeenCalled();
} finally {
setIntervalSpy.mockRestore();
}
});
it("does not replay hot-restart adoption when the orphan reaper retries", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
heartbeatSchedulerEnabled: true,
heartbeatSchedulerIntervalMs: 30000,
}));
heartbeatServiceMock.reconcileHotRestartAdoption.mockRejectedValueOnce(new Error("partial adoption"));
heartbeatServiceMock.reapOrphanedRuns
.mockRejectedValueOnce(new Error("transient reap failure"))
.mockResolvedValueOnce({ reaped: 0, runIds: [] });
await startServer();
expect(heartbeatServiceMock.reconcileHotRestartAdoption).toHaveBeenCalledTimes(1);
expect(heartbeatServiceMock.reapOrphanedRuns).toHaveBeenCalledTimes(2);
});
it("closes the bound listener when native startup recovery fails", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
heartbeatSchedulerEnabled: true,
heartbeatSchedulerIntervalMs: 30000,
}));
heartbeatServiceMock.recoverNativeRunsAfterRestart.mockRejectedValueOnce(
new Error("native recovery unavailable"),
);
await expect(startServer()).rejects.toThrow("native recovery unavailable");
expect(fakeServer.listen).toHaveBeenCalledTimes(1);
expect(fakeServer.close).toHaveBeenCalledTimes(1);
});
it("refuses authenticated public startup without an external database URL", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
deploymentExposure: "public",
authBaseUrlMode: "explicit",
authPublicBaseUrl: "https://tenant.example.com",
databaseMode: "embedded-postgres",
databaseUrl: undefined,
}));
await expect(startServer()).rejects.toThrow(
"authenticated public deployments require DATABASE_URL or config.database.connectionString",
);
expect(createDbMock).not.toHaveBeenCalled();
});
it("refuses authenticated public startup when DATABASE_URL is not a postgres URL", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
deploymentExposure: "public",
authBaseUrlMode: "explicit",
authPublicBaseUrl: "https://tenant.example.com",
databaseUrl: "secret://paperclip-cloud/stacks/alpha/database/runtime-url",
}));
await expect(startServer()).rejects.toThrow(
"authenticated public deployments require DATABASE_URL to be a postgres/postgresql connection string",
);
expect(createDbMock).not.toHaveBeenCalled();
});
});
describe("startServer authenticated auth origin setup", () => {
beforeEach(() => {
vi.clearAllMocks();
process.env.PAPERCLIP_DECISION_SIGNING_SECRET = "fedcba9876543210fedcba9876543210";
loadConfigMock.mockReturnValue(buildTestConfig());
createBetterAuthInstanceMock.mockReturnValue({});
deriveAuthTrustedOriginsMock.mockReturnValue([]);
process.env.BETTER_AUTH_SECRET = "test-secret";
});
it("checks port availability on the configured bind host", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
host: "127.0.0.1",
port: 3210,
}));
await startServer();
expect(detectPortMock).toHaveBeenCalledWith({
port: 3210,
hostname: "127.0.0.1",
});
});
it("derives trusted origins from the detected listen port before auth initializes", async () => {
loadConfigMock.mockReturnValue(buildTestConfig({
port: 3210,
allowedHostnames: ["board.example.test"],
authBaseUrlMode: "explicit",
authPublicBaseUrl: "http://127.0.0.1:3210",
}));
detectPortMock.mockResolvedValueOnce(3211);
deriveAuthTrustedOriginsMock.mockImplementation(
(_config: { port: number; authPublicBaseUrl?: string }, opts?: { listenPort?: number }) => [
`http://board.example.test:${opts?.listenPort ?? 0}`,
],
);
await startServer();
expect(deriveAuthTrustedOriginsMock).toHaveBeenCalledWith(
expect.objectContaining({
port: 3210,
authPublicBaseUrl: "http://127.0.0.1:3211/",
}),
{ listenPort: 3211 },
);
expect(createBetterAuthInstanceMock).toHaveBeenCalledWith(
expect.anything(),
expect.objectContaining({
port: 3210,
authPublicBaseUrl: "http://127.0.0.1:3211/",
}),
["http://board.example.test:3211"],
);
expect(createAppMock.mock.calls[0]?.[1]).toMatchObject({
serverPort: 3211,
});
});
});
describe("startServer PAPERCLIP_API_URL handling", () => {
beforeEach(() => {
vi.clearAllMocks();
process.env.PAPERCLIP_DECISION_SIGNING_SECRET = "fedcba9876543210fedcba9876543210";
loadConfigMock.mockReturnValue(buildTestConfig());
process.env.BETTER_AUTH_SECRET = "test-secret";
delete process.env.PAPERCLIP_API_URL;
});
afterEach(() => {
if (ORIGINAL_PAPERCLIP_API_URL === undefined) delete process.env.PAPERCLIP_API_URL;
else process.env.PAPERCLIP_API_URL = ORIGINAL_PAPERCLIP_API_URL;
if (ORIGINAL_PAPERCLIP_RUNTIME_API_URL === undefined) delete process.env.PAPERCLIP_RUNTIME_API_URL;
else process.env.PAPERCLIP_RUNTIME_API_URL = ORIGINAL_PAPERCLIP_RUNTIME_API_URL;
if (ORIGINAL_PAPERCLIP_RUNTIME_API_CANDIDATES_JSON === undefined) {
delete process.env.PAPERCLIP_RUNTIME_API_CANDIDATES_JSON;
} else {
process.env.PAPERCLIP_RUNTIME_API_CANDIDATES_JSON = ORIGINAL_PAPERCLIP_RUNTIME_API_CANDIDATES_JSON;
}
if (ORIGINAL_PAPERCLIP_LISTEN_HOST === undefined) delete process.env.PAPERCLIP_LISTEN_HOST;
else process.env.PAPERCLIP_LISTEN_HOST = ORIGINAL_PAPERCLIP_LISTEN_HOST;
if (ORIGINAL_PAPERCLIP_LISTEN_PORT === undefined) delete process.env.PAPERCLIP_LISTEN_PORT;
else process.env.PAPERCLIP_LISTEN_PORT = ORIGINAL_PAPERCLIP_LISTEN_PORT;
});
it("uses the externally set PAPERCLIP_API_URL when provided", async () => {
process.env.PAPERCLIP_API_URL = "http://custom-api:3100";
const started = await startServer();
expect(started.apiUrl).toBe("http://custom-api:3100");
expect(process.env.PAPERCLIP_API_URL).toBe("http://custom-api:3100");
expect(JSON.parse(process.env.PAPERCLIP_RUNTIME_API_CANDIDATES_JSON ?? "[]")).toEqual(
expect.arrayContaining(["http://custom-api:3100"]),
);
expect(JSON.parse(process.env.PAPERCLIP_RUNTIME_API_CANDIDATES_JSON ?? "[]")[0]).toBe("http://custom-api:3100");
});
it("falls back to host-based URL when PAPERCLIP_API_URL is not set", async () => {
const started = await startServer();
expect(started.apiUrl).toBe("http://127.0.0.1:3210");
expect(process.env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:3210");
});
it("keeps loopback as the runtime API URL when allowed hostnames are present", async () => {
loadConfigMock.mockReturnValueOnce(buildTestConfig({
allowedHostnames: ["192.168.1.50"],
}));
const started = await startServer();
expect(started.apiUrl).toBe("http://127.0.0.1:3210");
expect(process.env.PAPERCLIP_RUNTIME_API_URL).toBe("http://127.0.0.1:3210");
expect(process.env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:3210");
expect(JSON.parse(process.env.PAPERCLIP_RUNTIME_API_CANDIDATES_JSON ?? "[]")).toEqual(
expect.arrayContaining(["http://127.0.0.1:3210", "http://192.168.1.50:3210"]),
);
});
it("preserves explicit-port external auth public URLs when detect-port selects a new port", async () => {
loadConfigMock.mockReturnValueOnce(buildTestConfig({
port: 3100,
authBaseUrlMode: "explicit",
authPublicBaseUrl: "http://my-host.ts.net:3100",
}));
detectPortMock.mockResolvedValueOnce(3110);
const started = await startServer();
// The server listens internally on 3110, but an explicit *external* base URL must keep
// its advertised port. Rewriting it to the internal listen port produced an unreachable
// URL that leaked to spawned agents as a dead PAPERCLIP_API_URL. (BRO-1558)
expect(started.listenPort).toBe(3110);
expect(started.apiUrl).toBe("http://my-host.ts.net:3100");
expect(process.env.PAPERCLIP_RUNTIME_API_URL).toBe("http://my-host.ts.net:3100");
});
it("keeps no-port auth public URLs stable when detect-port selects a new port", async () => {
loadConfigMock.mockReturnValueOnce(buildTestConfig({
port: 3100,
authBaseUrlMode: "explicit",
authPublicBaseUrl: "https://paperclip.example",
}));
detectPortMock.mockResolvedValueOnce(3110);
const started = await startServer();
expect(started.listenPort).toBe(3110);
expect(started.apiUrl).toBe("https://paperclip.example");
expect(process.env.PAPERCLIP_RUNTIME_API_URL).toBe("https://paperclip.example");
});
});