mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-13 15:55:46 +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>
169 lines
8.7 KiB
Python
169 lines
8.7 KiB
Python
"""One explicitly authorized, owned initdb bootstrap diagnostic; no server/model/case.
|
|
Raw stderr remains private. Never stops unrelated processes or removes host IPC.
|
|
"""
|
|
import argparse
|
|
import datetime
|
|
import hashlib
|
|
import json
|
|
import os
|
|
import pathlib
|
|
import re
|
|
import secrets
|
|
import shutil
|
|
import signal
|
|
import stat
|
|
import subprocess
|
|
import tempfile
|
|
import time
|
|
|
|
SNAPSHOT = pathlib.Path('/private/tmp/paperclip-chat-candidate-0fdcb539a8')
|
|
PACKAGE = 'node_modules/.pnpm/@embedded-postgres+darwin-arm64@18.1.0-beta.16/node_modules/@embedded-postgres/darwin-arm64'
|
|
BINARY = SNAPSHOT / PACKAGE / 'native/bin/initdb'
|
|
PREFIX = 'paperclip-chat-initdb-diagnostic-'
|
|
|
|
|
|
def sha(filename):
|
|
return hashlib.sha256(pathlib.Path(filename).read_bytes()).hexdigest()
|
|
|
|
|
|
def private_json(filename, value):
|
|
with os.fdopen(os.open(filename, os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600), 'w') as output:
|
|
json.dump(value, output, indent=2)
|
|
output.write('\n')
|
|
|
|
|
|
def validate(binary, digest, root, token):
|
|
binary, root = pathlib.Path(binary), pathlib.Path(root)
|
|
if binary != BINARY or binary.resolve(strict=True) != BINARY or sha(binary) != digest:
|
|
raise ValueError('Unverified embedded initdb executable')
|
|
parent = pathlib.Path('/private/tmp').resolve()
|
|
if root.parent != parent or not root.name.startswith(PREFIX) or root.is_symlink():
|
|
raise ValueError('Wrong owned diagnostic root')
|
|
info = root.lstat()
|
|
if not stat.S_ISDIR(info.st_mode) or info.st_uid != os.getuid() or stat.S_IMODE(info.st_mode) != 0o700:
|
|
raise ValueError('Unsafe diagnostic root ownership')
|
|
marker = root / 'owner.json'
|
|
mark_info = marker.lstat()
|
|
if marker.is_symlink() or mark_info.st_uid != os.getuid() or stat.S_IMODE(mark_info.st_mode) != 0o600:
|
|
raise ValueError('Unsafe owner marker')
|
|
value = json.loads(marker.read_text())
|
|
if value != {'version': 1, 'creatorPid': os.getpid(), 'uid': os.getuid(), 'token': token}:
|
|
raise ValueError('Diagnostic ownership receipt mismatch')
|
|
return root / 'pgdata'
|
|
|
|
|
|
def group_members(group):
|
|
rows = []
|
|
output = subprocess.check_output(['ps', '-axo', 'uid=,pid=,ppid=,pgid=,comm='], text=True)
|
|
for line in output.splitlines():
|
|
parts = line.strip().split(None, 4)
|
|
if len(parts) == 5 and int(parts[3]) == group:
|
|
rows.append({'uid': int(parts[0]), 'pid': int(parts[1]), 'parentPid': int(parts[2]), 'group': int(parts[3]), 'executable': parts[4]})
|
|
return rows
|
|
|
|
|
|
def verify_members(rows):
|
|
for row in rows:
|
|
path = pathlib.Path(row['executable'])
|
|
if row['uid'] != os.getuid() or path.parent != BINARY.parent or path.name not in ['initdb', 'postgres']:
|
|
raise RuntimeError('Owned diagnostic process group cannot be safely verified')
|
|
|
|
|
|
def record_members(members, observed, ownership_findings):
|
|
observed.update({row['pid']: row for row in members})
|
|
try:
|
|
verify_members(members)
|
|
except RuntimeError:
|
|
ownership_findings.append('Process-group member failed strict executable ownership guard')
|
|
|
|
|
|
def cause(stderr, password):
|
|
lines = [line.replace(password, '[REDACTED]') for line in stderr.splitlines()
|
|
if re.search(r'FATAL:|PANIC:|ERROR:|initdb: error:|could not|No space left|Invalid argument', line, re.I)]
|
|
return lines[:10]
|
|
|
|
|
|
def run(expected_digest):
|
|
# Actual profile credentials and the campaign oracle key are never inherited.
|
|
original = SNAPSHOT.parent / 'paperclip-chat-candidate-e652b0de6e' / PACKAGE / 'native/bin/initdb'
|
|
if sha(original) != expected_digest or sha(BINARY) != expected_digest:
|
|
raise ValueError('Immutable dependency binary receipts differ')
|
|
version = subprocess.check_output([str(BINARY), '--version'], text=True, env={'PATH': '/usr/bin:/bin', 'LC_ALL': 'C'}).strip()
|
|
root = pathlib.Path(tempfile.mkdtemp(prefix=PREFIX, dir='/private/tmp'))
|
|
token = secrets.token_hex(24)
|
|
private_json(root / 'owner.json', {'version': 1, 'creatorPid': os.getpid(), 'uid': os.getuid(), 'token': token})
|
|
data = validate(BINARY, expected_digest, root, token)
|
|
if data.exists():
|
|
raise ValueError('Diagnostic cluster must not exist before admission')
|
|
password = secrets.token_urlsafe(32)
|
|
password_file = root / 'password'
|
|
with os.fdopen(os.open(password_file, os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600), 'w') as output:
|
|
output.write(password + '\n')
|
|
arguments = [str(BINARY), '--pgdata=' + str(data), '--auth=password', '--username=paperclip', '--pwfile=' + str(password_file), '--lc-messages=C', '--encoding=UTF8', '--locale=C', '--lc-messages=C']
|
|
started = datetime.datetime.now(datetime.timezone.utc).isoformat()
|
|
observed = {}
|
|
timed_out = False
|
|
ownership_findings = []
|
|
with os.fdopen(os.open(root / 'stdout.log', os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600), 'wb') as stdout, os.fdopen(os.open(root / 'stderr.log', os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600), 'wb') as stderr:
|
|
child = subprocess.Popen(arguments, stdout=stdout, stderr=stderr, start_new_session=True,
|
|
env={'PATH': '/usr/bin:/bin', 'LC_ALL': 'C', 'LC_MESSAGES': 'C', 'HOME': str(root), 'TMPDIR': str(root)})
|
|
print(json.dumps({"phase": "owned-bootstrap-started", "pid": child.pid, "privateRoot": str(root)}), flush=True)
|
|
deadline = time.monotonic() + 60
|
|
while child.poll() is None:
|
|
members = group_members(child.pid)
|
|
# Never lose stderr/status or signal an unverified transient member.
|
|
record_members(members, observed, ownership_findings)
|
|
if time.monotonic() > deadline:
|
|
timed_out = True
|
|
# The private session was created by this Popen. Verify all current
|
|
# members before either signal; never signal an unverified group.
|
|
try:
|
|
verify_members(group_members(child.pid))
|
|
except RuntimeError:
|
|
ownership_findings.append('Unverified group at deadline; only direct owned child may be stopped')
|
|
if ownership_findings:
|
|
child.terminate()
|
|
else:
|
|
os.killpg(child.pid, signal.SIGTERM)
|
|
try:
|
|
child.wait(timeout=5)
|
|
except subprocess.TimeoutExpired:
|
|
if ownership_findings:
|
|
child.kill()
|
|
else:
|
|
verify_members(group_members(child.pid))
|
|
os.killpg(child.pid, signal.SIGKILL)
|
|
break
|
|
time.sleep(0.05)
|
|
exit_code = child.wait(timeout=5)
|
|
remaining = group_members(child.pid)
|
|
if remaining:
|
|
ownership_findings.append('Diagnostic process group still has participants; cluster/password retained')
|
|
validate(BINARY, expected_digest, root, token)
|
|
stderr_text = (root / 'stderr.log').read_text(errors='replace')
|
|
# Only this freshly allocated marker-bound cluster is eligible for removal.
|
|
if data.is_symlink():
|
|
raise RuntimeError('Diagnostic pgdata became a symlink; root retained')
|
|
if not remaining:
|
|
if data.exists():
|
|
shutil.rmtree(data)
|
|
password_file.unlink()
|
|
receipt = {'schema': 'paperclip.initdb-infrastructure-diagnostic/v1', 'infrastructureOnly': True,
|
|
'canonicalAttempts': 0, 'providerCalls': 0, 'nativeRuns': 0, 'sourceSnapshot': str(SNAPSHOT),
|
|
'binary': str(BINARY), 'binarySha256': expected_digest, 'binaryVersion': version,
|
|
'helperSha256': sha(__file__), 'packageVersion': json.loads((SNAPSHOT / PACKAGE / 'package.json').read_text())['version'],
|
|
'command': arguments, 'startedAt': started, 'finishedAt': datetime.datetime.now(datetime.timezone.utc).isoformat(),
|
|
'pid': child.pid, 'processGroup': child.pid, 'observedOwnedProcesses': list(observed.values()),
|
|
'exitCode': exit_code, 'timedOut': timed_out, 'ownershipGuardFindings': sorted(set(ownership_findings)), 'stderrSha256': sha(root / 'stderr.log'),
|
|
'stdoutSha256': sha(root / 'stdout.log'), 'sanitizedCause': cause(stderr_text, password),
|
|
'cleanup': {'ownedProcessGroupEmpty': not remaining, 'ownedClusterDataRemoved': not data.exists(), 'passwordFileRemoved': not password_file.exists(), 'diagnosticLogsRetainedPrivately': True}}
|
|
private_json(root / 'receipt.json', receipt)
|
|
return {'receiptPath': str(root / 'receipt.json'), 'exitCode': exit_code, 'sanitizedCause': receipt['sanitizedCause'], 'cleanup': receipt['cleanup']}
|
|
|
|
|
|
if __name__ == '__main__':
|
|
parser = argparse.ArgumentParser(description=__doc__)
|
|
parser.add_argument('--expected-sha256', required=True)
|
|
args = parser.parse_args()
|
|
print(json.dumps(run(args.expected_sha256)))
|