RomanNum3ral
/
gstack
mirror of https://github.com/garrytan/gstack.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
gstack/browse/test
History
gus 01e584253d security: gate domain-skill auto-promote on classifier_score > 0 
`browse/src/domain-skill-commands.ts:140` (handleSave) writes
`classifier_score: 0` with the comment "L4 deferred to load-time / sidebar-agent
fills this in on first prompt-injection load." But CLAUDE.md "Sidebar
architecture" documents that sidebar-agent.ts was ripped, and grep for
recordSkillUse + classifierFlagged callers across browse/src/ returns zero hits
outside the module under test.

Net effect: every quarantined skill that survives three benign uses without
flag (`recordSkillUse(... , classifierFlagged: false)` x3) auto-promotes to
`active` and lands in prompt context wrapped as UNTRUSTED on every subsequent
visit to that host. The L4 score that was supposed to gate the promotion was
never written — the production save path puts 0 on disk and nothing later
updates it.

Threat model: a domain-skill body authored by an agent under the influence of
a poisoned page (the new `gstackInjectToTerminal` PTY path runs no L1-L3
either) would lose its auto-promote barrier after three uses. The exploit
isn't single-step but the bar is exactly N=3 prompt-injection-shaped uses on
a hostile page, which is well within reach.

Fix adds a single condition to the auto-promote gate in `recordSkillUse`:

    if (state === 'quarantined' && useCount >= PROMOTE_THRESHOLD &&
        flagCount === 0 && current.classifier_score > 0) {
      state = 'active';
    }

`classifier_score` is set once at writeSkill and never updated. Production
saves it as 0 (handleSave), so the gate stays closed; existing tests that
explicitly pass `classifierScore: 0.1` still auto-promote (the auto-promote
path is preserved for the day L4 is rewired).

Manual promotion via `domain-skill promote-to-global` is unaffected (it goes
through `promoteToGlobal` which has its own state-machine guard at line 337+).

Test: new regression case `does NOT auto-promote when classifier_score is 0
(production handleSave shape)` plants a skill with classifierScore=0 (matches
domain-skill-commands.ts:140), runs three uses without flag, asserts the skill
stays quarantined and readSkill returns null. Negative control: revert the
patch, the test fails with `Received: "active"`. With the patch: 15/15 pass.
 		2026-05-07 23:22:27 -03:00
..
fixtures v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
activity.test.ts feat: headed mode + sidebar agent + Chrome extension (v0.12.0) (#517) 2026-03-26 11:15:24 -06:00
adversarial-security.test.ts fix: security audit remediation — 12 fixes, 20 tests (v0.13.1.0) (#595) 2026-03-28 08:35:24 -06:00
batch.test.ts refactor: extract TabSession for per-tab state isolation (v0.15.16.0) (#873) 2026-04-07 00:23:36 -07:00
browse-client.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
browser-manager-unit.test.ts feat: headed mode + sidebar agent + Chrome extension (v0.12.0) (#517) 2026-03-26 11:15:24 -06:00
browser-skill-commands.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
browser-skill-write.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
browser-skills-e2e.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
browser-skills-storage.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
build.test.ts fix: ngrok Windows build + close CI error-swallowing gap (v0.18.0.1) (#1024) 2026-04-16 13:49:04 -07:00
bun-polyfill.test.ts fix: Windows support — Node.js server fallback for Playwright (#255) 2026-03-20 12:22:11 -07:00
cdp-allowlist.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
cdp-e2e.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
cdp-mutex.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
claude-bin.test.ts v1.24.0.0 feat: cross-platform hardening — curated Windows lane + Bun.which resolver + path-portability helper (#1252) 2026-05-01 07:21:28 -07:00
commands.test.ts feat(browse): Puppeteer parity — load-html, screenshot --selector, viewport --scale, file:// (v1.1.0.0) (#1062) 2026-04-18 23:25:33 +08:00
compare-board.test.ts refactor: extract TabSession for per-tab state isolation (v0.15.16.0) (#873) 2026-04-07 00:23:36 -07:00
config.test.ts fix: Windows browse — health-check-first ensureServer, detached startServer, Windows process mgmt (v0.11.11.0) (#431) 2026-03-24 00:38:10 -07:00
content-security.test.ts security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137) 2026-04-21 21:58:27 -07:00
cookie-import-browser.test.ts feat: Wave 3 — community bug fixes & platform support (v0.11.6.0) (#359) 2026-03-23 22:15:23 -07:00
cookie-picker-routes.test.ts community wave: 6 PRs + hardening (v0.18.1.0) (#1028) 2026-04-17 00:45:13 -07:00
data-platform.test.ts feat: browser data platform for AI agents (v0.16.0.0) (#907) 2026-04-08 00:41:55 -07:00
domain-skills-e2e.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
domain-skills-storage.test.ts security: gate domain-skill auto-promote on classifier_score > 0 2026-05-07 23:22:27 -03:00
dual-listener.test.ts v1.16.0.0 feat: tunnel allowlist 17→26 + canDispatchOverTunnel pure function (#1253) 2026-04-28 00:57:28 -07:00
dx-polish.test.ts feat(browse): Puppeteer parity — load-html, screenshot --selector, viewport --scale, file:// (v1.1.0.0) (#1062) 2026-04-18 23:25:33 +08:00
error-handling.test.ts refactor: AI slop reduction with cross-model quality review (v0.16.3.0) (#941) 2026-04-10 17:13:15 -10:00
file-drop.test.ts feat: headed mode + sidebar agent + Chrome extension (v0.12.0) (#517) 2026-03-26 11:15:24 -06:00
find-browse.test.ts feat: multi-agent support — gstack works on Codex, Gemini CLI, and Cursor (v0.9.0) (#226) 2026-03-19 18:20:50 -07:00
findport.test.ts feat: community PRs — faster install, skill namespacing, uninstall, Codex fallback, Windows fix, Python patterns (v0.12.9.0) (#561) 2026-03-27 00:44:37 -06:00
from-file-path-validation.test.ts security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137) 2026-04-21 21:58:27 -07:00
gstack-config.test.ts feat: composable skills — INVOKE_SKILL resolver + factoring infrastructure (v0.13.7.0) (#644) 2026-03-29 23:35:17 -06:00
gstack-update-check.test.ts fix: community PRs + security hardening + E2E stability (v0.12.7.0) (#552) 2026-03-26 23:21:27 -06:00
handoff.test.ts refactor: extract TabSession for per-tab state isolation (v0.15.16.0) (#873) 2026-04-07 00:23:36 -07:00
learnings-injection.test.ts fix: community security wave — 8 PRs, 4 contributors (v0.15.13.0) (#847) 2026-04-06 00:47:04 -07:00
pair-agent-e2e.test.ts security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137) 2026-04-21 21:58:27 -07:00
pair-agent-tunnel-eval.test.ts v1.16.0.0 feat: tunnel allowlist 17→26 + canDispatchOverTunnel pure function (#1253) 2026-04-28 00:57:28 -07:00
path-validation.test.ts fix: community security wave — 8 PRs, 4 contributors (v0.15.13.0) (#847) 2026-04-06 00:47:04 -07:00
pdf-flags.test.ts feat(v1.4.0.0): /make-pdf — markdown to publication-quality PDFs (#1086) 2026-04-20 13:20:30 +08:00
platform.test.ts fix: Windows support — Node.js server fallback for Playwright (#255) 2026-03-20 12:22:11 -07:00
security-adversarial-fixes.test.ts feat: gstack browser sidebar = interactive Claude Code REPL with live tab awareness (v1.14.0.0) (#1216) 2026-04-25 22:52:15 -07:00
security-adversarial.test.ts v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
security-audit-r2.test.ts feat: gstack browser sidebar = interactive Claude Code REPL with live tab awareness (v1.14.0.0) (#1216) 2026-04-25 22:52:15 -07:00
security-bench-ensemble-live.test.ts v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
security-bench-ensemble.test.ts v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
security-bench.test.ts v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
security-bunnative.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
security-classifier.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
security-integration.test.ts v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
security-live-playwright.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
security-review-flow.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
security-sidepanel-dom.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
security-source-contracts.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
security.test.ts v1.6.4.0: cut Haiku classifier FP from 44% to 23%, gate now enforced (#1135) 2026-04-23 10:23:40 -07:00
server-auth.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
sidebar-integration.test.ts security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137) 2026-04-21 21:58:27 -07:00
sidebar-security.test.ts feat(security): ML prompt injection defense for sidebar (v1.4.0.0) (#1089) 2026-04-20 22:18:37 +08:00
sidebar-tabs.test.ts feat: gstack browser sidebar = interactive Claude Code REPL with live tab awareness (v1.14.0.0) (#1216) 2026-04-25 22:52:15 -07:00
sidebar-unit.test.ts fix: sidebar agent uses real tab URL instead of stale Playwright URL (v0.12.6.0) (#544) 2026-03-26 22:07:03 -06:00
sidebar-ux.test.ts fix: community security wave — 8 PRs, 4 contributors (v0.15.13.0) (#847) 2026-04-06 00:47:04 -07:00
skill-token.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
snapshot.test.ts refactor: extract TabSession for per-tab state isolation (v0.15.16.0) (#873) 2026-04-07 00:23:36 -07:00
sse-session-cookie.test.ts security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137) 2026-04-21 21:58:27 -07:00
state-ttl.test.ts fix: security audit remediation — 12 fixes, 20 tests (v0.13.1.0) (#595) 2026-03-28 08:35:24 -06:00
tab-each.test.ts feat: gstack browser sidebar = interactive Claude Code REPL with live tab awareness (v1.14.0.0) (#1216) 2026-04-25 22:52:15 -07:00
tab-isolation.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
telemetry.test.ts v1.20.0.0 feat: browser-skills runtime + gbrain-support carryover (#1233) 2026-04-28 20:08:04 -07:00
terminal-agent-integration.test.ts feat: gstack browser sidebar = interactive Claude Code REPL with live tab awareness (v1.14.0.0) (#1216) 2026-04-25 22:52:15 -07:00
terminal-agent.test.ts feat: gstack browser sidebar = interactive Claude Code REPL with live tab awareness (v1.14.0.0) (#1216) 2026-04-25 22:52:15 -07:00
test-server.ts feat: Phase 3.5 — cookie import, QA testing, team retro (v0.3.1) (#29) 2026-03-13 00:31:41 -07:00
token-registry.test.ts feat: browser data platform for AI agents (v0.16.0.0) (#907) 2026-04-08 00:41:55 -07:00
tunnel-gate-unit.test.ts v1.16.0.0 feat: tunnel allowlist 17→26 + canDispatchOverTunnel pure function (#1253) 2026-04-28 00:57:28 -07:00
url-validation.test.ts security: tunnel dual-listener + SSRF + envelope + path wave (v1.6.0.0) (#1137) 2026-04-21 21:58:27 -07:00
watch.test.ts feat: headed mode + sidebar agent + Chrome extension (v0.12.0) (#517) 2026-03-26 11:15:24 -06:00
watchdog.test.ts community wave: 6 PRs + hardening (v0.18.1.0) (#1028) 2026-04-17 00:45:13 -07:00
welcome-page.test.ts feat: GStack Browser — double-click AI browser with anti-bot stealth (#695) 2026-04-04 10:17:05 -07:00