ci: run native checks on combined validation branch

.github/scripts/check_test_execution.py

@@ -0,0 +1,17 @@
+"""Reject skipped/empty native CI runs without conflating driver child counts."""
+import argparse
+from pathlib import Path
+import xml.etree.ElementTree as ET
+
+parser = argparse.ArgumentParser(description=__doc__)
+parser.add_argument('report', type=Path)
+parser.add_argument('--minimum', type=int, required=True)
+args = parser.parse_args()
+root = ET.parse(args.report).getroot()
+cases = list(root.iter('testcase'))
+if len(cases) < args.minimum:
+    raise SystemExit(f'Incomplete execution: {len(cases)} cases, expected at least {args.minimum}')
+for case in cases:
+    if any(case.find(tag) is not None for tag in ('skipped', 'failure', 'error')):
+        raise SystemExit(f'Case did not pass: {case.get("classname")}::{case.get("name")}')
+print(f'Verified {len(cases)} executed cases; no skipped, failed or errored cases')

.github/scripts/run_native_tests.mjs

@@ -0,0 +1,46 @@
+import fs from 'node:fs';
+import path from 'node:path';
+import { fileURLToPath } from 'node:url';
+import { spawnSync } from 'node:child_process';
+import { parse } from 'acorn';
+
+const root = fileURLToPath(new URL('../../', import.meta.url));
+const manifest = JSON.parse(fs.readFileSync(path.join(root, 'tests/native-ci-manifest.json'), 'utf8'));
+function files(directory) {
+  return fs.readdirSync(directory, { withFileTypes: true }).flatMap(entry => {
+    const target = path.join(directory, entry.name);
+    return entry.isDirectory() ? files(target) : /\.(?:mjs|cjs|js)$/.test(entry.name) ? [target] : [];
+  });
+}
+function importsNativeTest(node) {
+  if (!node || typeof node !== 'object') return false;
+  if (node.type === 'ImportDeclaration' && node.source.value === 'node:test') return true;
+  if (node.type === 'CallExpression' && node.callee.name === 'require' && node.arguments[0]?.value === 'node:test') return true;
+  return Object.values(node).some(value => Array.isArray(value) ? value.some(importsNativeTest) : importsNativeTest(value));
+}
+const discovered = files(path.join(root, 'tests')).filter(file => {
+  const source = fs.readFileSync(file, 'utf8');
+  return source.includes('node:test') && importsNativeTest(parse(source, { ecmaVersion: 'latest', sourceType: 'module' }));
+}).map(file => path.relative(root, file).split(path.sep).join('/')).sort();
+const assigned = Object.values(manifest).flatMap(group => Array.isArray(group) ? group : Object.keys(group));
+if (new Set(assigned).size !== assigned.length || JSON.stringify([...assigned].sort()) !== JSON.stringify(discovered)) {
+  throw new Error('Native test coverage must be exhaustive and disjoint; update tests/native-ci-manifest.json for added, removed or duplicated files');
+}
+for (const wrappers of Object.values(manifest['python-ci'])) {
+  for (const wrapper of wrappers) if (!fs.existsSync(path.join(root, wrapper))) throw new Error(`Missing pytest wrapper: ${wrapper}`);
+}
+if (process.argv.includes('--list')) {
+  console.log(JSON.stringify({ discovered, runners: manifest }, null, 2));
+} else {
+  const reports = path.join(root, 'test-results/native-node');
+  fs.mkdirSync(reports, { recursive: true });
+  const result = spawnSync(process.execPath, ['--test', '--test-concurrency=1',
+    '--test-reporter=spec', '--test-reporter-destination=stdout',
+    '--test-reporter=junit', `--test-reporter-destination=${path.join(reports, 'junit.xml')}`,
+    ...manifest['native-ci']], {
+    cwd: root, stdio: 'inherit',
+    env: { ...process.env, ENDPOINT_URL: 'http://127.0.0.1:1' },
+  });
+  if (result.error) throw result.error;
+  process.exitCode = result.status ?? 1;
+}

.github/workflows/ci.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches: [main, dev]
   pull_request:
+  workflow_dispatch:
 
 # Least privilege: none of the jobs write to the repo.
 permissions:

.github/workflows/native-test-suites.yml

@@ -0,0 +1,119 @@
+name: Native test suites
+
+on:
+  pull_request:
+  push:
+    branches: [main, dev, test/remediation-integration-validation]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: native-tests-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  native-node:
+    name: Native Node regressions
+    runs-on: ubuntu-24.04
+    timeout-minutes: 15
+    steps:
+      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          persist-credentials: false
+      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: '3.11'
+      - uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
+        with:
+          node-version: '22.23.4'
+          cache: npm
+      - run: npm ci --ignore-scripts --no-audit --no-fund
+      - uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+        with:
+          path: ~/.cache/ms-playwright
+          key: ms-playwright-${{ runner.os }}-${{ hashFiles('package-lock.json') }}
+      - run: npx playwright install --with-deps chromium
+      - name: Validate exhaustive disjoint runner mapping and execute omitted suites
+        run: |
+          node .github/scripts/run_native_tests.mjs
+          python .github/scripts/check_test_execution.py test-results/native-node/junit.xml --minimum 43
+      - name: Save native test identities and results
+        if: always()
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: native-node-results
+          path: test-results/native-node/
+          retention-days: 7
+          if-no-files-found: warn
+
+  browser:
+    name: Chromium application regressions
+    runs-on: ubuntu-24.04
+    timeout-minutes: 30
+    env:
+      ODYSSEUS_E2E_DISPOSABLE_ARTIFACTS: '1'
+      ODYSSEUS_TEST_PYTHON: python
+      CHROMADB_HOST: 127.0.0.1
+      CHROMADB_PORT: '1'
+    steps:
+      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+        with:
+          persist-credentials: false
+      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: '3.11'
+      - name: Install FFmpeg and establish system trust
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends ffmpeg ca-certificates
+          python scripts/check_ca_trust.py --system
+      - uses: astral-sh/setup-uv@c18668ad3cf93ea998bef934396af7bb5c839dc7 # v10.2.0
+        with:
+          version: '0.13.5'
+          enable-cache: true
+          cache-dependency-glob: requirements.txt
+          prune-cache: false
+      - run: |
+          uv pip install --system --require-hashes --only-binary=:all: --no-deps -r requirements-bootstrap.txt
+          python scripts/check_ca_trust.py --python
+          uv pip install --system -r requirements.txt
+      - uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
+        with:
+          node-version: '22.23.4'
+          cache: npm
+      - run: npm ci --ignore-scripts --no-audit --no-fund
+      - uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+        with:
+          path: ~/.cache/ms-playwright
+          key: ms-playwright-${{ runner.os }}-${{ hashFiles('package-lock.json') }}
+      - run: npx playwright install --with-deps chromium
+      - name: Run complete Chromium suite against a fresh authenticated app
+        env:
+          PLAYWRIGHT_JUNIT_OUTPUT_FILE: test-results/browser-junit.xml
+        run: |
+          npm run test:photo-editor -- --reporter=line,junit
+          python .github/scripts/check_test_execution.py test-results/browser-junit.xml --minimum 130
+      - name: Save per-test results and sanitized failure diagnostics
+        if: always()
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: chromium-regression-results
+          path: test-results/
+          retention-days: 7
+          if-no-files-found: warn
+
+  authoritative-native-suites:
+    name: Authoritative native suites
+    if: always()
+    needs: [native-node, browser]
+    runs-on: ubuntu-24.04
+    steps:
+      - name: Require both complete suites to succeed
+        env:
+          NODE_RESULT: ${{ needs.native-node.result }}
+          BROWSER_RESULT: ${{ needs.browser.result }}
+        run: |
+          test "$NODE_RESULT" = success
+          test "$BROWSER_RESULT" = success

docs/maintainers/native-test-coverage.md

@@ -0,0 +1,38 @@
+# Native test merge gate
+
+`Native test suites` runs on pull requests, dev/main pushes and manual dispatch.
+`Authoritative native suites` fails unless both the native Node and complete
+Chromium jobs succeed; it has no continue-on-error, retries or path bypass.
+Repository administrators must require this check alongside the four pytest
+checks in branch protection. The audited public dev branch had no protection or
+ruleset, so merely adding workflow jobs cannot prevent a manual merge.
+
+`tests/native-ci-manifest.json` exhaustively partitions all native Node files:
+16 run in the native job,14 through existing pytest wrappers, one through the
+existing frontend architecture CI step, and the formerly live email-reply file
+through an explicitly identified Playwright driver. The manifest validator fails
+for omitted/duplicate/unreachable entries. Node emits individual JUnit identities
+and propagates its failure exit. No native file is newly run in each pytest shard.
+
+The browser job uses the existing one-worker Chromium config and all original129
+cases, plus one driver for the native email-reply contract. Its ephemeral account,
+SQLite DB and server are created by the config/setup; reuseExistingServer remains
+false. NPX startup is offline/cache-only in a private cache; Notes/editor behavior
+and Python MCP services remain real. No developer app or external account is used.
+
+Default CI trace suppression remains intact. Only this job's explicit opt-in plus
+a newly generated authenticated synthetic fixture marker permits failure traces.
+The existing structured sanitizer runs before artifacts are uploaded; invalid
+ownership, failed sanitization and missing tooling purge sensitive artifacts.
+Never upload the fixture directory, database or browser storage state.
+
+The stroke budget now excludes measured synthetic event-delivery pacing and also
+bounds actual Chromium main-thread work at the original5s. All240 samples and
+pixel/thumbnail/undo/redo assertions remain. Per-case timing attachments expose
+both clocks rather than treating driver frame cadence as application work.
+
+Three Bombadil properties remain opt-in: they require a Bombadil/browser scenario
+and bounded fuzz campaign; no authoritative execution is claimed. The two model
+smoke scripts require supplied model endpoints/models and remain explicit live
+operations. No credentials or live-service authorization are implied by CI setup.
+Firefox/WebKit alternatives remain opt-in and are not certified by Chromium CI.

scripts/verify_notes_browser_runtime.py

@@ -11,6 +11,38 @@ import time
 from urllib.request import urlopen
 
 ROOT = Path(__file__).resolve().parents[1]
+if str(ROOT) not in sys.path:
+    sys.path.insert(0, str(ROOT))
+
+
+def _stop_owned_server(server, identity):
+    """Close only identity-bound descendants, including MCP's new sessions.
+
+    Capture before signalling the parent: an orphaned stderr holder can no
+    longer be attributed by ancestry. Never search by executable name or kill
+    the caller's process group.
+    """
+    from src import process_lifecycle as lifecycle
+
+    descendants = (lifecycle.bind_descendants([identity.pid])
+                   if identity.verdict() == lifecycle.OWNED else [])
+    if os.name == "nt":
+        if identity.verdict() == lifecycle.OWNED:
+            lifecycle.taskkill_tree(identity.pid)
+    else:
+        outcome = lifecycle.terminate_identities(
+            [*(entry.identity for entry in reversed(descendants)), identity],
+            steps=lifecycle.term_kill_steps(10),
+        )
+        if outcome.survivors or outcome.unverified:
+            raise RuntimeError(f"Disposable Notes cleanup could not verify exit: {outcome}")
+    try:
+        logs, _ = server.communicate(timeout=2)
+    except subprocess.TimeoutExpired as error:
+        # Do not turn an unidentified pipe holder into an unbounded wait.
+        server.stdout.close()
+        raise RuntimeError("Disposable Notes output pipe remained open after owned cleanup") from error
+    return logs
 
 
 def main():
@@ -36,7 +68,10 @@ def main():
         server = subprocess.Popen(
             [sys.executable, "-m", "uvicorn", "app:app", "--host", "127.0.0.1", "--port", str(port)],
             cwd=target, env=env, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True,
+            start_new_session=(os.name != "nt"),
         )
+        from src.process_lifecycle import ProcessIdentity
+        server_identity = ProcessIdentity.capture(server.pid)
         result = None
         try:
             origin = f"http://127.0.0.1:{port}"
@@ -62,12 +97,7 @@ def main():
                 print(result.stderr, file=sys.stderr)
             return result.returncode
         finally:
-            server.terminate()
-            try:
-                logs, _ = server.communicate(timeout=10)
-            except subprocess.TimeoutExpired:
-                server.kill()
-                logs, _ = server.communicate()
+            logs = _stop_owned_server(server, server_identity)
             if result is None or result.returncode:
                 print(logs[-8000:], file=sys.stderr)
 

tests/TESTING_STANDARD.md

@@ -99,8 +99,10 @@ cleanup. Specifically:
   `os.environ[...] = ...` that outlives the test.
 - **Current working directory** - never `chdir` without restoring; never assert
   against cwd-relative paths like `./data`. Use a temp workspace helper instead.
-- **Database** - the root `conftest.py` defaults `DATABASE_URL` to an in-memory
-  SQLite for collection safety. A test that needs a real file-backed DB must opt
+- **Database** - the root `conftest.py` defaults `DATABASE_URL` to a private,
+  worker-owned SQLite file for collection safety and consistent schema across
+  TestClient threads. This fallback is not a per-test data fixture. A test that
+  writes or depends on particular rows must opt
   in explicitly via `tests.helpers.sqlite_db.make_temp_sqlite` and bind its
   `SessionLocal` onto the module under test. Do not rely on a persistent
   on-disk DB existing.

tests/conftest.py

@@ -12,9 +12,17 @@ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 # File-backed databases remain fixture-owned. Cleanup restores the caller.
 # Keep the existing generated environment-reference locations stable.
 _database_environment = pytest.MonkeyPatch()
-_database_environment.setenv("DATABASE_URL", "sqlite:///:memory:")
 from tests.helpers.worker_runtime import bootstrap_runtime, configure_runtime
 _runtime_environment = bootstrap_runtime()
+# A file keeps the collection schema consistent across TestClient threads;
+# SingletonThreadPool can evict the populated in-memory main-thread database.
+# One patcher owns this variable, including restoration of an inherited URL.
+_database_environment.setenv(
+    "DATABASE_URL",
+    f"sqlite:///{_runtime_environment.root / 'collection.db'}"
+    if _runtime_environment is not None else "sqlite:///:memory:",
+)
+_collection_database_engine = None
 
 # Pre-import real heavy modules BEFORE any test file's module-level stubs can
 # replace them with MagicMock. Some test files (e.g. test_llm_core_sanitize_*)
@@ -26,6 +34,7 @@ try:
     import sqlalchemy.orm  # noqa: F401
     import core.database  # noqa: F401
     import src.database
+    _collection_database_engine = core.database.engine
 except ImportError:
     pass  # not installed - the stubs below will handle it
 
@@ -100,6 +109,10 @@ def pytest_configure(config):
     only registers marker names; it imports no production module.
     """
     config.add_cleanup(_database_environment.undo)
+    if _collection_database_engine is not None:
+        # Registered after runtime cleanup, so connections close before the
+        # owned file is removed (also required on Windows).
+        config.add_cleanup(_collection_database_engine.dispose)
 
     import pathlib
     from tests._taxonomy import discover_markers

tests/e2e/artifact-policy.js

@@ -0,0 +1,19 @@
+const fs = require('node:fs');
+const os = require('node:os');
+const path = require('node:path');
+const marker = '.disposable-ci-fixture.json';
+
+function disposableCIArtifacts() {
+  if (!(process.env.CI || process.env.GITHUB_ACTIONS) || process.env.ODYSSEUS_E2E_DISPOSABLE_ARTIFACTS !== '1') return false;
+  const directory = process.env.PHOTO_EDITOR_E2E_DATA_DIR;
+  if (!directory || process.env.PHOTO_EDITOR_E2E_AUTH === '0') return false;
+  const resolved = path.resolve(directory);
+  if (!path.basename(resolved).startsWith('odysseus-photo-editor-e2e-')) return false;
+  if (![os.tmpdir(), '/tmp', '/var/tmp'].some(base => resolved.startsWith(path.resolve(base) + path.sep))) return false;
+  if (process.env.PHOTO_EDITOR_E2E_DB_PATH !== path.join(resolved, 'app.db')) return false;
+  try {
+    const fixture = JSON.parse(fs.readFileSync(path.join(resolved, marker), 'utf8'));
+    return fixture.synthetic === true && fixture.authenticated === true && fixture.database === path.join(resolved, 'app.db');
+  } catch { return false; }
+}
+module.exports = { disposableCIArtifacts, marker };

tests/e2e/photo-editor/brush-engine.spec.js

@@ -35,7 +35,7 @@ test('brush presets and shared stroke controls remain reusable', async ({ page }
   await expect(page.locator('#ge-brush-preset option', { hasText: 'My Detail Brush' })).toHaveCount(0);
 });
 
-test('long sampled strokes stay responsive and undo atomically', async ({ page }) => {
+test('long sampled strokes stay responsive and undo atomically', async ({ page, browserName }, testInfo) => {
   await openBlankEditor(page);
   await selectTool(page, 'brush');
   await page.locator('.ge-size-slider').fill('700');
@@ -49,13 +49,40 @@ test('long sampled strokes stay responsive and undo atomically', async ({ page }
   const before = await flattenedPixelDigest(page);
   const start = await canvasPoint(page, 0.08, 0.5);
   const end = await canvasPoint(page, 0.92, 0.5);
+  // Playwright delivers stepped moves at frame cadence: 240 events already
+  // cost ~4s on an inert page. Measure that transport cost with the same input,
+  // and separately bound real Chromium main-thread work at the original5s.
+  const control = await page.context().newPage();
+  let controlDuration;
+  try {
+    await control.setContent('<canvas style="position:fixed;inset:0;width:100vw;height:100vh"></canvas>');
+    const started = Date.now();
+    await control.mouse.move(start.x, start.y);
+    await control.mouse.down();
+    await control.mouse.move(end.x, end.y, { steps: 240 });
+    await control.mouse.up();
+    controlDuration = Date.now() - started;
+  } finally {
+    await control.close();
+    await page.bringToFront();
+  }
+  const metrics = browserName === 'chromium' ? await page.context().newCDPSession(page) : null;
+  if (metrics) await metrics.send('Performance.enable');
+  const taskTime = async () => (await metrics.send('Performance.getMetrics')).metrics.find(item => item.name === 'TaskDuration').value *1000;
+  const beforeTask = metrics ? await taskTime() : null;
   const startedAt = Date.now();
   await page.mouse.move(start.x, start.y);
   await page.mouse.down();
   await page.mouse.move(end.x, end.y, { steps: 240 });
   await page.mouse.up();
   const strokeDuration = Date.now() - startedAt;
-  expect(strokeDuration).toBeLessThan(5000);
+  const mainThreadDuration = metrics ? (await taskTime()) - beforeTask : null;
+  if (metrics) await metrics.detach();
+  await testInfo.attach('stroke-timing', { body: Buffer.from(JSON.stringify({
+    samples: 240, strokeDuration, controlDuration, mainThreadDuration,
+  })), contentType: 'application/json' });
+  expect(strokeDuration - controlDuration).toBeLessThan(5000);
+  if (mainThreadDuration !== null) expect(mainThreadDuration).toBeLessThan(5000);
   const after = await flattenedPixelDigest(page);
   expect(after).not.toEqual(before);
   await expect(thumb).toHaveAttribute('width', '68');

tests/e2e/photo-editor/email-reply-native.spec.js

@@ -0,0 +1,21 @@
+const { test, expect } = require('@playwright/test');
+const { spawnSync } = require('node:child_process');
+const path = require('node:path');
+
+// The previously opt-in node:test now uses only this config's owned app/account.
+// Its native JUnit report retains its authored identity; this case is a driver.
+test('native email reply regression on the disposable authenticated app', async ({ context, baseURL }, testInfo) => {
+  const origin = new URL(baseURL);
+  expect(origin.hostname).toBe('127.0.0.1');
+  const cookie = (await context.cookies(baseURL)).find(item => item.name === 'odysseus_session');
+  expect(cookie?.value).toBeTruthy();
+  const report = testInfo.outputPath('native-email-junit.xml');
+  const result = spawnSync(process.execPath, ['--test', '--test-reporter=spec', '--test-reporter-destination=stdout',
+    '--test-reporter=junit', `--test-reporter-destination=${report}`, 'tests/emailReplyBrowser.test.mjs'], {
+    cwd: path.resolve(__dirname, '../../..'), encoding: 'utf8', timeout: 80_000,
+    env: { ...process.env, ODYSSEUS_UI_TEST_URL: baseURL, ODYSSEUS_UI_TEST_COOKIE: cookie.value },
+  });
+  console.log(result.stdout);
+  expect(result.error, String(result.error)).toBeUndefined();
+  expect(result.status, result.stdout + result.stderr).toBe(0);
+});

tests/e2e/playwright.config.js

@@ -2,6 +2,12 @@ const fs = require('node:fs');
 const os = require('node:os');
 const path = require('node:path');
 const { defineConfig } = require('@playwright/test');
+const { disposableCIArtifacts, marker } = require('./artifact-policy');
+const isCI = Boolean(process.env.CI) || Boolean(process.env.GITHUB_ACTIONS);
+const disposableOptIn = isCI && process.env.ODYSSEUS_E2E_DISPOSABLE_ARTIFACTS === '1';
+if (disposableOptIn && (process.env.PHOTO_EDITOR_E2E_DATA_DIR || process.env.PHOTO_EDITOR_E2E_DB_PATH || process.env.PHOTO_EDITOR_E2E_AUTH === '0') && !disposableCIArtifacts()) {
+  throw new Error('CI artifact opt-in requires fresh authenticated fixture paths');
+}
 
 const port = Number(process.env.PHOTO_EDITOR_E2E_PORT || 7013);
 const baseURL = `http://127.0.0.1:${port}`;
@@ -24,7 +30,9 @@ if (!['chromium', 'firefox', 'webkit'].includes(browserName)) {
   throw new Error(`Unsupported PHOTO_EDITOR_E2E_BROWSER: ${browserName}`);
 }
 
-const isCI = Boolean(process.env.CI) || Boolean(process.env.GITHUB_ACTIONS);
+if (disposableOptIn) {
+  fs.writeFileSync(path.join(dataDirectory, marker), JSON.stringify({ synthetic: true, authenticated: true, database: databasePath }));
+}
 const traceOptIn = !isCI && ['1', 'true'].includes(String(process.env.PLAYWRIGHT_ENABLE_TRACE || '').toLowerCase());
 
 module.exports = defineConfig({
@@ -32,6 +40,8 @@ module.exports = defineConfig({
   outputDir: path.join(__dirname, '..', '..', 'test-results', 'photo-editor'),
   timeout: 90_000,
   expect: { timeout: 10_000 },
+  forbidOnly: isCI,
+  retries: 0,
   fullyParallel: false,
   workers: 1,
   reporter: isCI ? 'line' : 'line',
@@ -49,9 +59,9 @@ module.exports = defineConfig({
     serviceWorkers: 'block',
     ...(authenticated ? { storageState: path.join(dataDirectory, 'browser-auth.json') } : {}),
     screenshot: 'only-on-failure',
-    // Unconditional CI trace suppression: Playwright traces are NEVER recorded in CI.
+    // Default CI suppression remains; only the fresh synthetic fixture opts in.
     // Local developer tracing requires explicit PLAYWRIGHT_ENABLE_TRACE=1.
-    trace: traceOptIn ? 'retain-on-failure' : 'off',
+    trace: (traceOptIn || disposableCIArtifacts()) ? 'retain-on-failure' : 'off',
   },
   webServer: {
     command: `${JSON.stringify(python)} -m uvicorn app:app --host 127.0.0.1 --port ${port}`,
@@ -65,6 +75,10 @@ module.exports = defineConfig({
       DATABASE_URL: `sqlite:///${databasePath}`,
       ODYSSEUS_DATA_DIR: dataDirectory,
       ODYSSEUS_STARTUP_WARMUPS: '0',
+      ODYSSEUS_TOOL_INDEX_PREWARM: '0',
+      ODYSSEUS_BROWSER_MCP_REQUIRE_CACHE: '1',
+      npm_config_offline: 'true',
+      npm_config_cache: path.join(dataDirectory, 'npm-cache'),
       RESPONSE_COMPRESSION_ENABLED: 'false',
     },
   },

tests/e2e/teardown.js

@@ -2,6 +2,7 @@ const fs = require('node:fs');
 const os = require('node:os');
 const path = require('node:path');
 const { execFileSync } = require('node:child_process');
+const { disposableCIArtifacts } = require('./artifact-policy');
 
 const repoRoot = path.resolve(__dirname, '..', '..');
 const defaultResultsDir = path.join(repoRoot, 'test-results');
@@ -29,6 +30,9 @@ module.exports = async (options = {}) => {
   const isPlaywrightConfig = options && (options.configFile || options.rootDir);
   const testOpts = (!isPlaywrightConfig && typeof options === 'object') ? options : {};
 
+  // Validate fixture ownership before removing its temporary directory.
+  const disposableCI = disposableCIArtifacts();
+
   // Confine cleanup of PHOTO_EDITOR_E2E_DATA_DIR:
   // Strictly require it to be within an OS temporary directory and match the expected test prefix
   const dataDirectory = process.env.PHOTO_EDITOR_E2E_DATA_DIR;
@@ -100,7 +104,7 @@ module.exports = async (options = {}) => {
     const isCI = Boolean(process.env.CI) || Boolean(process.env.GITHUB_ACTIONS);
     const traceOptIn = !isCI && ['1', 'true'].includes(String(process.env.PLAYWRIGHT_ENABLE_TRACE || '').toLowerCase());
 
-    if (isCI || !traceOptIn) {
+    if ((isCI && !disposableCI) || (!traceOptIn && !disposableCI)) {
       // In CI or default local runs, purge all sensitive trace archives and artifacts
       purgeSensitiveArtifacts(resultsDir);
     } else {
@@ -124,7 +128,8 @@ module.exports = async (options = {}) => {
 
       if (!sanitizerSucceeded) {
         // Fail closed: purge all remaining zip or tmp archives
-        purgeZipAndTmp(resultsDir);
+        if (disposableCI) purgeSensitiveArtifacts(resultsDir);
+        else purgeZipAndTmp(resultsDir);
       }
     }
   }

tests/native-ci-manifest.json

@@ -0,0 +1,71 @@
+{
+  "native-ci": [
+    "tests/backgroundToolJobs.test.mjs",
+    "tests/chatEditorProgress.test.mjs",
+    "tests/chatImageDeletion.test.mjs",
+    "tests/chatPhotoEditorIntegrity.test.mjs",
+    "tests/chatProcessingHandoff.test.mjs",
+    "tests/copyChatHeader.test.mjs",
+    "tests/emailReplyStream.test.mjs",
+    "tests/galleryModelPickerSync.test.mjs",
+    "tests/generatedImageResult.test.mjs",
+    "tests/researchMobileButtons.test.mjs",
+    "tests/schemaThinkingProbe.test.mjs",
+    "tests/sidebarNewChat.test.mjs",
+    "tests/skillsApproval.test.mjs",
+    "tests/streamingRendererRaf.test.mjs",
+    "tests/toolFollowupOracle.test.mjs",
+    "tests/tool_followup_oracle.test.mjs"
+  ],
+  "browser-ci": [
+    "tests/emailReplyBrowser.test.mjs"
+  ],
+  "python-ci": {
+    "tests/historyResumeRendering.test.mjs": [
+      "tests/test_history_resume_rendering_js.py",
+      "tests/test_live_fallback_round_attribution.py"
+    ],
+    "tests/live_thinking_scheduler.test.mjs": [
+      "tests/test_live_thinking_scheduler_js.py"
+    ],
+    "tests/noteTestOracle.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesApplication.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesCharacterization.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesDraftAutosave.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesInfrastructure.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesRecurrence.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesReminders.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/notesWorkflows.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/selectedNotesContext.test.mjs": [
+      "tests/test_notes_reference_contracts_js.py"
+    ],
+    "tests/streaming/invariant.test.mjs": [
+      "tests/test_streaming_segmenter_js.py"
+    ],
+    "tests/streaming/segmenter.test.mjs": [
+      "tests/test_streaming_segmenter_js.py"
+    ],
+    "tests/turnRendering.test.mjs": [
+      "tests/test_turn_rendering_js.py"
+    ]
+  },
+  "architecture-ci": [
+    "tests/frontendArchitecture.test.mjs"
+  ]
+}

tests/test_agent_editor_routes.py

@@ -1929,23 +1929,60 @@ def test_place_layer_reports_false_on_queued_response(monkeypatch):
     assert placed is False
 
 
-def test_bridge_await_cancellation_handled_cleanly():
+def test_bridge_await_cancellation_handled_cleanly(monkeypatch):
     import asyncio
+    from contextlib import suppress
     from httpx import ASGITransport, AsyncClient
     app = _bridge_app()
 
     async def scenario():
+        entered_wait = asyncio.Event()
+        real_wait_for = asyncio.wait_for
+        waits = []
+
+        async def observe_wait(awaitable, timeout):
+            # Observe the real wait; do not replace its result or cancellation.
+            waits.append(awaitable)
+            entered_wait.set()
+            return await real_wait_for(awaitable, timeout)
+
+        monkeypatch.setattr(eaar.asyncio, "wait_for", observe_wait)
         async with AsyncClient(transport=ASGITransport(app=app), base_url="http://test") as c:
             cmd_resp = await c.post("/api/editor/bridge/command", json={"session": "cancel-sess", "op": "document.info", "wait": False})
+            assert cmd_resp.status_code == 200
+            assert cmd_resp.json()["completed"] is False
             cid = cmd_resp.json()["queued"][0]
 
             await_task = asyncio.create_task(c.post("/api/editor/bridge/await", json={"id": cid, "timeout": 30}))
-            await asyncio.sleep(0.05)
-            await_task.cancel()
             try:
-                await await_task
-            except asyncio.CancelledError:
-                pass
+                await real_wait_for(entered_wait.wait(), 5)
+                assert len(waits) == 1
+                assert not waits[0].done()
+                assert not await_task.done(), "await returned without a queued result"
+                await_task.cancel()
+                response = await real_wait_for(await_task, 5)
+                assert response.status_code == 200
+                assert response.json() == {
+                    "ok": False, "error": "cancelled", "status": "cancelled",
+                    "completed": False, "id": cid,
+                }
+            finally:
+                if not await_task.done():
+                    await_task.cancel()
+                    with suppress(asyncio.CancelledError):
+                        await await_task
+
+            # Cancelling the waiter must not cancel the queued command itself.
+            payload = {"width": 137, "height": 251}
+            delivered = await c.post("/api/editor/bridge/result", json={"id": cid, "ok": True, "result": payload})
+            assert delivered.status_code == 200
+            completed = await c.post("/api/editor/bridge/await", json={"id": cid, "timeout": 1})
+            assert completed.status_code == 200
+            assert completed.json()["status"] == "completed"
+            assert completed.json()["completed"] is True
+            assert completed.json()["ok"] is True
+            assert completed.json()["result"] == payload
+            assert len(waits) == 2
 
     asyncio.run(scenario())
 

tests/test_auth_config_lock_concurrency.py

@@ -206,35 +206,50 @@ class TestDiskConsistency:
     @pytest.mark.slow
     def test_file_always_valid_json_during_concurrent_ops(self, tmp_path):
         mgr = _fresh_auth_manager(tmp_path)
-        mgr.create_user("admin", "adminpw", is_admin=True)
+        assert mgr.create_user("admin", "adminpw", is_admin=True)
 
         stop_event = threading.Event()
+        reader_ready = threading.Event()
         corruption_found = []
+        observations = {"reads": 0, "with_writes": 0}
 
         def reader():
             while not stop_event.is_set():
                 try:
                     with open(mgr.auth_path, "r") as f:
                         content = f.read()
-                    json.loads(content)
+                    data = json.loads(content)
+                    observations["reads"] += 1
+                    if any(name.startswith("stress") for name in data["users"]):
+                        observations["with_writes"] += 1
+                    reader_ready.set()
                 except json.JSONDecodeError as e:
                     corruption_found.append(str(e))
                     break
-                except FileNotFoundError:
-                    pass
                 time.sleep(0.001)
 
         def writer():
+            assert reader_ready.wait(5), "reader never observed the initial file"
+            results = []
             for i in range(50):
-                mgr.create_user(f"stress{i}", f"pw{i}")
+                results.append(mgr.create_user(f"stress{i}", f"pw{i}"))
+            return results
 
-        reader_thread = threading.Thread(target=reader)
-        writer_thread = threading.Thread(target=writer)
-
-        reader_thread.start()
-        writer_thread.start()
-        writer_thread.join()
-        stop_event.set()
-        reader_thread.join()
+        # Futures propagate reader/writer exceptions to the test thread.
+        with ThreadPoolExecutor(max_workers=2) as pool:
+            reading = pool.submit(reader)
+            writing = pool.submit(writer)
+            try:
+                results = writing.result(timeout=120)
+            finally:
+                stop_event.set()
+            reading.result(timeout=5)
 
         assert not corruption_found, f"Corrupt JSON detected: {corruption_found[0]}"
+        assert len(results) == 50 and all(results), "writer operations did not succeed"
+        assert observations["reads"] > 0
+        assert observations["with_writes"] > 0, "reader saw no persisted writer activity"
+        expected = {"admin"} | {f"stress{i}" for i in range(50)}
+        with open(mgr.auth_path) as persisted:
+            assert set(json.load(persisted)["users"]) == expected
+        assert set(_fresh_auth_manager(tmp_path).users) == expected

tests/test_database_test_isolation.py

@@ -85,6 +85,9 @@ def test_actual_tests_restore_process_state_and_ignore_inherited_database(tmp_pa
         import os
         import sys
         import pytest
+        import tempfile
+        from pathlib import Path
+        from sqlalchemy.engine import make_url
 
         def snapshot():
             import core
@@ -107,7 +110,11 @@ def test_actual_tests_restore_process_state_and_ignore_inherited_database(tmp_pa
         class StateGuard:
             def pytest_sessionstart(self):
                 self.initial = snapshot()
-                assert self.initial[0] == "sqlite:///:memory:"
+                assert self.initial[0] != inherited_url
+                self.default_path = Path(make_url(self.initial[0]).database)
+                assert self.default_path.is_file()
+                assert self.default_path.name == "collection.db"
+                assert self.default_path.parent == Path(tempfile.gettempdir()).parent
 
             def pytest_collection_finish(self):
                 assert snapshot() == self.initial, "collection changed database bindings"
@@ -118,9 +125,11 @@ def test_actual_tests_restore_process_state_and_ignore_inherited_database(tmp_pa
                 assert snapshot() == self.initial, "test leaked database or module state"
 
         inherited_url = os.environ["DATABASE_URL"]
+        guard = StateGuard()
         result = pytest.main(["-q", "-p", "no:cacheprovider", *sys.argv[1:]],
-                             plugins=[StateGuard()])
+                             plugins=[guard])
         assert os.environ["DATABASE_URL"] == inherited_url
+        assert not guard.default_path.exists(), "collection database outlived its runtime"
         raise SystemExit(result)
     ''')
     result = subprocess.run(

tests/test_delete_user_invalidates_token_cache.py

@@ -11,6 +11,8 @@ cache from the DB, where the rows are already gone, and the token is rejected.
 """
 import asyncio
 import types
+import pytest
+from fastapi import HTTPException
 
 from routes.auth_routes import setup_auth_routes, DeleteUserRequest
 
@@ -60,12 +62,9 @@ def test_refused_delete_does_not_invalidate_cache():
     invalidations = []
     router = setup_auth_routes(_auth_manager(delete_result=False))
     handler = _handler(router)
-    try:
+    with pytest.raises(HTTPException) as refused:
         asyncio.run(handler(DeleteUserRequest(username="admin"), _fake_request(invalidations)))
-        raised = False
-    except Exception:
-        raised = True
-    assert raised, "a refused delete should raise (HTTP 400)"
+    assert refused.value.status_code == 400
     assert invalidations == [], "a refused delete must not touch the token cache"
 
 

tests/test_editor_photoshop_features_more4.py

@@ -93,7 +93,7 @@ def test_custom_filter_reuses_convolve_kernel():
     assert "convolve(readSource(snap), k, divisor, bias, 1)" in EFFECTS
 
 
-def test_new_filter_aliases_registered():
+def test_blur_and_custom_filter_aliases_registered():
     assert "'box-blur': 'blur-box'" in EDITOR
     assert "'surface-blur': 'blur-surface'" in EDITOR
     assert "custom: 'other-custom'" in EDITOR

tests/test_email_list_stale_while_revalidate.py

@@ -56,14 +56,35 @@ def test_cold_start_with_no_snapshot_and_no_index_is_not_marked_stale(monkeypatc
     # falls through to the real (blocking) path, so the response is not a
     # stale-marked snapshot.
     endpoint = _list_endpoint()
+    calls = []
+
+    class EmptyMailbox:
+        def select(self, folder, readonly=False):
+            calls.append(("select", folder, readonly))
+            return "OK", [b"0"]
+
+        def uid(self, command, *args):
+            assert command == "SEARCH"
+            calls.append(("search", args))
+            return "OK", [b""]
+
+        def logout(self):
+            pass
+
+    def connect(*args, **kwargs):
+        calls.append(("connect", args, kwargs))
+        return EmptyMailbox()
+
+    monkeypatch.setattr(er, "_start_poller", lambda: None)
     monkeypatch.setattr(er, "_email_index_list", lambda *a, **k: ([], 0, None))
-    monkeypatch.setattr(er, "_imap_connect", lambda *a, **k: (_ for _ in ()).throw(
-        RuntimeError("offline test")))
-    try:
-        result = _call(endpoint)
-    except Exception:
-        return  # the cold path legitimately attempted real IMAP
+    monkeypatch.setattr(er, "_imap_connect", connect)
+    result = _call(endpoint)
+    assert [call[0] for call in calls] == ["connect", "select", "search"]
+    assert calls[1][2] is True
     assert isinstance(result, dict)
+    assert result["emails"] == []
+    assert result["total"] == 0
+    assert not result.get("error")
     assert not (result.get("sync") or {}).get("stale")
 
 

tests/test_notes_native_cleanup.py

@@ -0,0 +1,43 @@
+"""A detached Notes descendant must not hold output open or kill a bystander."""
+import os
+import runpy
+import subprocess
+import sys
+from pathlib import Path
+
+import pytest
+from src import process_lifecycle as lifecycle
+
+ROOT = Path(__file__).resolve().parents[1]
+
+
+@pytest.mark.skipif(os.name == "nt", reason="POSIX detached-session regression; Windows uses taskkill tree")
+def test_notes_cleanup_reaps_detached_pipe_holder_and_preserves_bystander():
+    stop = runpy.run_path(str(ROOT / "scripts/verify_notes_browser_runtime.py"))["_stop_owned_server"]
+    child = "import time; time.sleep(60)"
+    program = ("import subprocess,sys,time; "
+               f"p=subprocess.Popen([sys.executable,'-c',{child!r}],start_new_session=True); "
+               "print(p.pid,flush=True); time.sleep(60)")
+    with subprocess.Popen([sys.executable, "-c", child]) as bystander:
+        server = subprocess.Popen([sys.executable, "-c", program], stdout=subprocess.PIPE,
+                                  stderr=subprocess.STDOUT, text=True, start_new_session=True)
+        root_identity = lifecycle.ProcessIdentity.capture(server.pid)
+        detached = None
+        try:
+            child_pid = int(server.stdout.readline())
+            detached = lifecycle.ProcessIdentity.capture(child_pid)
+            assert os.getpgid(child_pid) == child_pid
+            stop(server, root_identity)
+            assert server.poll() is not None
+            assert detached.exited()
+            assert bystander.poll() is None, "cleanup touched an unrelated process"
+        finally:
+            # Test-owned fallback cleanup also runs when the negative control
+            # intentionally substitutes the old parent-only teardown.
+            targets = [identity for identity in (detached, root_identity) if identity]
+            lifecycle.terminate_identities(targets, steps=lifecycle.term_kill_steps(1))
+            server.wait(timeout=3)
+            if server.stdout:
+                server.stdout.close()
+            bystander.terminate()
+            bystander.wait(timeout=3)

tests/test_playwright_teardown_security.py

@@ -303,3 +303,52 @@ def test_teardown_rejects_unconfined_programmatic_paths():
     res = _run_teardown(options={"resultsDir": "/etc"})
     assert res.returncode == 1
     assert "Security rejection" in res.stderr
+
+
+def _disposable_ci_fixture(tmp_path):
+    data = tmp_path / "odysseus-photo-editor-e2e-fixture"
+    data.mkdir()
+    database = data / "app.db"
+    (data / ".disposable-ci-fixture.json").write_text(json.dumps({
+        "synthetic": True, "authenticated": True, "database": str(database),
+    }))
+    return {"CI": "true", "GITHUB_ACTIONS": "true",
+            "ODYSSEUS_E2E_DISPOSABLE_ARTIFACTS": "1",
+            "PHOTO_EDITOR_E2E_DATA_DIR": str(data), "PHOTO_EDITOR_E2E_DB_PATH": str(database),
+            "PHOTO_EDITOR_E2E_AUTH": "1", "ODYSSEUS_TEST_PYTHON": sys.executable}
+
+
+def test_disposable_ci_retains_only_sanitized_failure_artifacts(tmp_path):
+    results = tmp_path / "results"
+    results.mkdir()
+    trace = results / "trace.zip"
+    with zipfile.ZipFile(trace, "w") as archive:
+        archive.writestr("0-trace.trace", json.dumps({"type": "action", "name": "fill", "params": {"value": "synthetic-secret"}}) + "\n")
+    screenshot = results / "failure.png"
+    screenshot.write_bytes(b"synthetic screenshot")
+    response = _run_teardown(_disposable_ci_fixture(tmp_path), {"resultsDir": str(results)})
+    assert response.returncode == 0, response.stderr
+    with zipfile.ZipFile(trace) as archive:
+        assert json.loads(archive.read("0-trace.trace"))["params"]["value"] == "[REDACTED]"
+    assert screenshot.exists()
+
+
+@pytest.mark.parametrize("defect", ["missing-marker", "wrong-database", "unauthenticated", "sanitizer-failure"])
+def test_disposable_ci_artifact_failures_purge_closed(tmp_path, defect):
+    results = tmp_path / "results"
+    results.mkdir()
+    (results / "trace.zip").write_bytes(b"invalid sensitive archive")
+    (results / "failure.png").write_bytes(b"sensitive screenshot")
+    environment = _disposable_ci_fixture(tmp_path)
+    if defect == "missing-marker":
+        (Path(environment["PHOTO_EDITOR_E2E_DATA_DIR"]) / ".disposable-ci-fixture.json").unlink()
+    elif defect == "wrong-database":
+        environment["PHOTO_EDITOR_E2E_DB_PATH"] = str(tmp_path / "other.db")
+    elif defect == "unauthenticated":
+        environment["PHOTO_EDITOR_E2E_AUTH"] = "0"
+    else:
+        environment["ODYSSEUS_TEST_PYTHON"] = "/missing-python"
+    response = _run_teardown(environment, {"resultsDir": str(results)})
+    assert response.returncode == 0, response.stderr
+    assert not (results / "trace.zip").exists()
+    assert not (results / "failure.png").exists()

tests/test_procfs_scan_guard.py

@@ -12,6 +12,8 @@ exercising every procfs-touching code path on both a Linux and a non-Linux
 host. ``test_cookbook_stop_without_procfs.py`` covers the behaviour itself.
 """
 import ast
+
+import pytest
 from pathlib import Path
 
 REPO_ROOT = Path(__file__).resolve().parent.parent
@@ -75,37 +77,112 @@ def _scan_sites(tree: ast.AST) -> list[ast.Call]:
     return sites
 
 
-def _guard_lines(tree: ast.AST) -> list[int]:
-    """Line numbers of calls that test whether procfs is present."""
-    lines = []
-    for node in ast.walk(tree):
-        if not isinstance(node, ast.Call):
-            continue
-        func = node.func
-        name = func.attr if isinstance(func, ast.Attribute) else getattr(func, "id", "")
-        if name not in _GUARD_FUNCS:
-            continue
-        subject_is_procfs = (
-            name in {"has_procfs", "is_wsl"}
-            or (isinstance(func, ast.Attribute) and _is_procfs_root(func.value))
-            or any(_is_procfs_root(arg) for arg in node.args)
-        )
-        if subject_is_procfs:
-            lines.append(node.lineno)
-    return lines
+def _condition_requires_procfs(node: ast.AST, truth: bool) -> bool:
+    """Whether this Boolean outcome proves a direct procfs check succeeded.
+
+    An ignored call, another function's check, or a check on only one possible
+    OR branch provides no proof. Keep unknown expressions conservative.
+    """
+    if isinstance(node, ast.UnaryOp) and isinstance(node.op, ast.Not):
+        return _condition_requires_procfs(node.operand, not truth)
+    if isinstance(node, ast.BoolOp):
+        facts = [_condition_requires_procfs(value, truth) for value in node.values]
+        all_values_required = isinstance(node.op, ast.And) == truth
+        return any(facts) if all_values_required else all(facts)
+    if not isinstance(node, ast.Call):
+        return False
+    func = node.func
+    name = func.attr if isinstance(func, ast.Attribute) else getattr(func, "id", "")
+    subject_is_procfs = (
+        name == "has_procfs"
+        or (isinstance(func, ast.Attribute) and _is_procfs_root(func.value))
+        or any(_is_procfs_root(arg) for arg in node.args)
+    )
+    return truth and name in _GUARD_FUNCS and subject_is_procfs
 
 
-def _enclosing_scope(tree: ast.AST, node: ast.AST) -> ast.AST:
-    """Smallest function/module scope containing ``node``."""
-    best = tree
-    for candidate in ast.walk(tree):
-        if not isinstance(candidate, (ast.FunctionDef, ast.AsyncFunctionDef)):
-            continue
-        end = getattr(candidate, "end_lineno", None) or candidate.lineno
-        if candidate.lineno <= node.lineno <= end:
-            if best is tree or candidate.lineno > best.lineno:
-                best = candidate
-    return best
+def _guarded_scan_ids(tree: ast.AST) -> set[int]:
+    """Follow conditional/early-exit paths instead of comparing line numbers.
+
+    Each function starts with no procfs assumption. Facts from a loop or try
+    body cannot escape to code which may run without that body completing.
+    This intentionally requires direct checks; arbitrary aliases/call graphs
+    are not a proof that procfs exists at the enumeration site.
+    """
+    scans = {id(site) for site in _scan_sites(tree)}
+    guarded = set()
+
+    def expression(node, present):
+        if node is None:
+            return
+        if id(node) in scans and present:
+            guarded.add(id(node))
+        if isinstance(node, ast.BoolOp):
+            for value in node.values:
+                expression(value, present)
+                # Later AND operands run only after a true prefix; later OR
+                # operands run only after a false prefix.
+                present = present or _condition_requires_procfs(value, isinstance(node.op, ast.And))
+        elif isinstance(node, ast.IfExp):
+            expression(node.test, present)
+            expression(node.body, present or _condition_requires_procfs(node.test, True))
+            expression(node.orelse, present or _condition_requires_procfs(node.test, False))
+        elif isinstance(node, (ast.ListComp, ast.SetComp, ast.GeneratorExp, ast.DictComp)):
+            for generator in node.generators:
+                expression(generator.iter, present)
+                for condition in generator.ifs:
+                    expression(condition, present)
+                    present = present or _condition_requires_procfs(condition, True)
+            for attr in ("elt", "key", "value"):
+                expression(getattr(node, attr, None), present)
+        elif isinstance(node, ast.Lambda):
+            expression(node.body, False)
+        else:
+            for child in ast.iter_child_nodes(node):
+                expression(child, present)
+
+    def block(statements, present):
+        for statement in statements:
+            if isinstance(statement, (ast.FunctionDef, ast.AsyncFunctionDef)):
+                block(statement.body, False)
+                for value in statement.decorator_list + statement.args.defaults:
+                    expression(value, present)
+            elif isinstance(statement, ast.ClassDef):
+                block(statement.body, present)
+            elif isinstance(statement, ast.If):
+                expression(statement.test, present)
+                yes = block(statement.body, present or _condition_requires_procfs(statement.test, True))
+                no = block(statement.orelse, present or _condition_requires_procfs(statement.test, False))
+                continuing = [fact for fact in (yes, no) if fact is not None]
+                if not continuing:
+                    return None
+                present = all(continuing)
+            elif isinstance(statement, (ast.Return, ast.Raise, ast.Break, ast.Continue)):
+                for child in ast.iter_child_nodes(statement):
+                    expression(child, present)
+                return None
+            elif isinstance(statement, (ast.For, ast.AsyncFor, ast.While)):
+                condition = statement.test if isinstance(statement, ast.While) else statement.iter
+                expression(condition, present)
+                body_present = present or (isinstance(statement, ast.While) and _condition_requires_procfs(condition, True))
+                block(statement.body, body_present)
+                block(statement.orelse, present)
+            elif isinstance(statement, (ast.Try, getattr(ast, "TryStar", ast.Try))):
+                block(statement.body, present)
+                block(statement.orelse, present)
+                block(statement.finalbody, present)
+                for handler in statement.handlers:
+                    block(handler.body, present)
+            elif isinstance(statement, (ast.With, ast.AsyncWith)):
+                for item in statement.items:
+                    expression(item.context_expr, present)
+                block(statement.body, present)
+            else:
+                expression(statement, present)
+        return present
+
+    block(tree.body, False)
+    return guarded
 
 
 def _app_python_files() -> list[Path]:
@@ -139,12 +216,10 @@ def _collect_sites() -> tuple[list[str], list[str]]:
         sites = _scan_sites(tree)
         if not sites:
             continue
-        guards = _guard_lines(tree)
+        guards = _guarded_scan_ids(tree)
         for site in sites:
-            scope = _enclosing_scope(tree, site)
-            start = getattr(scope, "lineno", 0)
             label = f"{path.relative_to(REPO_ROOT)}:{site.lineno}"
-            if any(start <= g < site.lineno for g in guards):
+            if id(site) in guards:
                 guarded.append(label)
             else:
                 unguarded.append(label)
@@ -154,7 +229,7 @@ def _collect_sites() -> tuple[list[str], list[str]]:
 def test_every_procfs_scan_is_guarded_by_an_existence_check():
     _guarded, unguarded = _collect_sites()
     assert not unguarded, (
-        "procfs pid scans with no existence check in the enclosing function — "
+        "procfs pid scans without an effective conditional existence guard — "
         "these raise FileNotFoundError on macOS and Windows: "
         + ", ".join(unguarded)
     )
@@ -176,3 +251,49 @@ def test_the_guard_detector_still_sees_the_known_scans():
         f"root was renamed. Found: {sorted(found)}"
     )
     assert len(found) >= 3, f"expected at least 3 procfs scans, found {sorted(found)}"
+
+
+@pytest.mark.parametrize("body", [
+    'Path("/proc").exists(); os.listdir("/proc")',
+    'if Path("/proc").exists():\n    pass\nos.listdir("/proc")',
+    'if not Path("/proc").exists():\n    os.listdir("/proc")',
+    'if ready or Path("/proc").exists():\n    os.listdir("/proc")',
+    'if ready:\n    Path("/proc").exists()\nos.listdir("/proc")',
+    'os.listdir("/proc"); Path("/proc").exists()',
+    'if Path("/tmp").exists():\n    os.listdir("/proc")',
+    'def other():\n    Path("/proc").exists()\nos.listdir("/proc")',
+    'if not has_procfs():\n    pass\nos.listdir("/proc")',
+    'for x in []:\n    if not has_procfs():\n        return\nos.listdir("/proc")',
+    'if has_procfs():\n    return\nelse:\n    os.listdir("/proc")',
+    '[pid for pid in os.listdir("/proc") if has_procfs()]',
+], ids=["ignored", "expired-branch", "inverted", "unsafe-or", "sibling-branch",
+        "too-late", "wrong-path", "other-function", "no-early-exit", "loop-may-not-run",
+        "wrong-else", "late-comprehension-filter"])
+def test_guard_detector_rejects_ineffective_checks(body):
+    tree = ast.parse("def scan():\n" + "\n".join("    " + line for line in body.splitlines()))
+    sites = _scan_sites(tree)
+    assert sites, "negative control must contain a recognized enumeration"
+    assert any(id(site) not in _guarded_scan_ids(tree) for site in sites)
+
+
+@pytest.mark.parametrize("body", [
+    'if Path("/proc").exists():\n    os.listdir("/proc")',
+    'if not Path("/proc").exists():\n    return\nos.listdir("/proc")',
+    'if not has_procfs():\n    raise RuntimeError()\nos.listdir(PROC_ROOT)',
+    'if has_procfs():\n    pass\nelse:\n    return\nPROC_ROOT.iterdir()',
+    'if ready and has_procfs():\n    os.scandir("/proc")',
+    'if not has_procfs() or disabled:\n    return\nPROC_ROOT.glob("*")',
+    'if sys.platform == "linux" and has_procfs():\n    PROC_ROOT.iterdir()',
+    'for item in items:\n    if not has_procfs():\n        continue\n    os.listdir("/proc")',
+    'if os.path.isdir("/proc"):\n    os.listdir("/proc")',
+    'if has_procfs() and list(PROC_ROOT.iterdir()):\n    pass',
+    'if not has_procfs() or list(PROC_ROOT.iterdir()):\n    pass',
+    '[pid for item in items if has_procfs() for pid in os.listdir("/proc")]',
+], ids=["positive-branch", "early-return", "early-raise", "else-exits", "and",
+        "negative-or", "platform-and-procfs", "loop-continue", "isdir",
+        "short-circuit-and", "short-circuit-or", "guarded-comprehension"])
+def test_guard_detector_preserves_effective_checks(body):
+    tree = ast.parse("def scan():\n" + "\n".join("    " + line for line in body.splitlines()))
+    sites = _scan_sites(tree)
+    assert sites, "positive control must contain a recognized enumeration"
+    assert all(id(site) in _guarded_scan_ids(tree) for site in sites)

tests/test_worker_runtime.py

@@ -79,3 +79,32 @@ def test_tmp_path_under_private_runtime_fits_unix_socket(tmp_path):
     # witness. A nested pytest-of-<user> basetemp made it 110 bytes under xdist.
     with socket.socket(socket.AF_UNIX) as sock:
         sock.bind(str(tmp_path / "tmux.sock"))
+
+
+def test_collection_database_survives_thread_connection_churn():
+    """TestClient threads must not lose the collection database or its schema.
+
+    SingletonThreadPool's default five-connection limit used to close the main
+    thread's in-memory database after repeated route clients. Each new thread
+    also saw an empty database. Exercise more threads than that limit and then
+    verify both compatibility factories still see all registered tables.
+    """
+    from concurrent.futures import ThreadPoolExecutor
+    from sqlalchemy import inspect
+    import core.database as database
+    import src.database as compatibility
+
+    expected = set(database.Base.metadata.tables)
+
+    def table_names(factory):
+        with factory() as session:
+            return set(inspect(session.connection()).get_table_names())
+
+    assert {"documents", "model_endpoints", "mcp_servers", "calendars"} <= expected
+    assert expected <= table_names(database.SessionLocal)
+    for _ in range(8):
+        # A fresh executor creates a fresh thread rather than reusing one.
+        with ThreadPoolExecutor(max_workers=1) as executor:
+            assert expected <= executor.submit(table_names, database.SessionLocal).result()
+    assert expected <= table_names(database.SessionLocal)
+    assert expected <= table_names(compatibility.SessionLocal)