Commit Graph

5 Commits

Author SHA1 Message Date
Garry Tan 7e96fe299b fix: security wave 3 — 12 fixes, 7 contributors (v0.16.4.0) (#988)
* fix(security): validateOutputPath symlink bypass — check file-level symlinks

validateOutputPath() previously only resolved symlinks on the parent directory.
A symlink at /tmp/evil.png → /etc/crontab passed the parent check (parent is
/tmp, which is safe) but the write followed the symlink outside safe dirs.

Add lstatSync() check: if the target file exists and is a symlink, resolve
through it and verify the real target is within SAFE_DIRECTORIES. ENOENT
(file doesn't exist yet) falls through to the existing parent-dir check.

Closes #921

Co-Authored-By: Yunsu <Hybirdss@users.noreply.github.com>

* fix(security): shell injection in bin/ scripts — use env vars instead of interpolation

gstack-settings-hook interpolated $SETTINGS_FILE directly into bun -e
double-quoted blocks. A path containing quotes or backticks breaks the JS
string context, enabling arbitrary code execution.

Replace direct interpolation with environment variables (process.env).
Same fix applied to gstack-team-init which had the same pattern.

Systematic audit confirmed only these two scripts were vulnerable — all
other bin/ scripts already use stdin piping or env vars.

Closes #858

Co-Authored-By: Gus <garagon@users.noreply.github.com>

* fix(security): cookie-import path validation bypass + hardcoded /tmp

Two fixes:
1. cookie-import relative path bypass (#707): path.isAbsolute() gated the
   entire validation, so relative paths like "sensitive-file.json" bypassed
   the safe-directory check entirely. Now always resolves to absolute path
   with realpathSync for symlink resolution, matching validateOutputPath().

2. Hardcoded /tmp in cookie-import-browser (#708): openDbFromCopy used
   /tmp directly instead of os.tmpdir(), breaking Windows support.

Also adds explicit imports for SAFE_DIRECTORIES and isPathWithin in
write-commands.ts (previously resolved implicitly through bundler).

Closes #852

Co-Authored-By: Toby Morning <urbantech@users.noreply.github.com>

* fix(security): redact form fields with sensitive names, not just type=password

Form redaction only applied to type="password" fields. Hidden and text
fields named csrf_token, api_key, session_id, etc. were exposed unredacted
in LLM context, leaking secrets.

Extend redaction to check field name and id against sensitive patterns:
token, secret, key, password, credential, auth, jwt, session, csrf, sid,
api_key. Uses the same pattern style as SENSITIVE_COOKIE_NAME.

Closes #860

Co-Authored-By: Gus <garagon@users.noreply.github.com>

* fix(security): restrict session file permissions to owner-only

Design session files written to /tmp with default umask (0644) were
world-readable on shared systems. Sessions contain design prompts and
feedback history.

Set mode 0o600 (owner read/write only) on both create and update paths.

Closes #859

Co-Authored-By: Gus <garagon@users.noreply.github.com>

* fix(security): enforce frozen lockfile during setup

bun install without --frozen-lockfile resolves ^semver ranges from npm on
every run. If an attacker publishes a compromised compatible version of any
dependency, the next ./setup pulls it silently.

Add --frozen-lockfile with fallback to plain install (for fresh clones
where bun.lock may not exist yet). Matches the pattern already used in
the .agents/ generation block (line 237).

Closes #614

Co-Authored-By: Alberto Martinez <halbert04@users.noreply.github.com>

* fix: remove duplicate recursive chmod on /tmp in Dockerfile.ci

chmod -R 1777 /tmp recursively sets sticky bit on files (no defined
behavior), not just the directory. Deduplicate to single chmod 1777 /tmp.

Closes #747

Co-Authored-By: Maksim Soltan <Gonzih@users.noreply.github.com>

* fix(security): learnings input validation + cross-project trust gate

Three fixes to the learnings system:

1. Input validation in gstack-learnings-log: type must be from allowed list,
   key must be alphanumeric, confidence must be 1-10 integer, source must
   be from allowed list. Prevents injection via malformed fields.

2. Prompt injection defense: insight field checked against 10 instruction-like
   patterns (ignore previous, system:, override, etc.). Rejected with clear
   error message.

3. Cross-project trust gate in gstack-learnings-search: AI-generated learnings
   from other projects are filtered out. Only user-stated learnings cross
   project boundaries. Prevents silent prompt injection across codebases.

Also adds trusted field (true for user-stated source, false for AI-generated)
to enable the trust gate at read time.

Closes #841

Co-Authored-By: Ziad Al Sharif <Ziadstr@users.noreply.github.com>

* feat(security): track cookie-imported domains and scope cookie imports

Foundation for origin-pinned JS execution (#616). Tracks which domains
cookies were imported from so the JS/eval commands can verify execution
stays within imported origins.

Changes:
- BrowserManager: new cookieImportedDomains Set with track/get/has methods
- cookie-import: tracks imported cookie domains after addCookies
- cookie-import-browser: tracks domains on --domain direct import
- cookie-import-browser --all: new explicit opt-in for all-domain import
  (previously implicit behavior, now requires deliberate flag)

Closes #615

Co-Authored-By: Alberto Martinez <halbert04@users.noreply.github.com>

* feat(security): pin JS/eval execution to cookie-imported origins

When cookies have been imported for specific domains, block JS execution
on pages whose origin doesn't match. Prevents the attack chain:
1. Agent imports cookies for github.com
2. Prompt injection navigates to attacker.com
3. Agent runs js document.cookie → exfiltrates github cookies

assertJsOriginAllowed() checks the current page hostname against imported
cookie domains with subdomain matching (.github.com allows api.github.com).
When no cookies are imported, all origins allowed (nothing to protect).
about:blank and data: URIs are allowed (no cookies at risk).

Depends on #615 (cookie domain tracking).

Closes #616

Co-Authored-By: Alberto Martinez <halbert04@users.noreply.github.com>

* feat(security): add persistent command audit log

Append-only JSONL audit trail for all browse server commands. Unlike
in-memory ring buffers, the audit log persists across restarts and is
never truncated. Each entry records: timestamp, command, args (truncated
to 200 chars), page origin, duration, status, error (truncated to 300
chars), hasCookies flag, connection mode.

All writes are best-effort — audit failures never block command execution.
Log stored at ~/.gstack/.browse/browse-audit.jsonl.

Closes #617

Co-Authored-By: Alberto Martinez <halbert04@users.noreply.github.com>

* fix(security): block hex-encoded IPv4-mapped IPv6 metadata bypass

URL constructor normalizes ::ffff:169.254.169.254 to ::ffff:a9fe:a9fe
(hex form), which was not in the blocklist. Similarly, ::169.254.169.254
normalizes to ::a9fe:a9fe.

Add both hex-encoded forms to BLOCKED_METADATA_HOSTS so they're caught
by the direct hostname check in validateNavigationUrl.

Closes #739

Co-Authored-By: Osman Mehmood <mehmoodosman@users.noreply.github.com>

* chore: bump version and changelog (v0.16.4.0)

Security wave 3: 12 fixes, 7 contributors.
Cookie origin pinning, command audit log, domain tracking.
Symlink bypass, path validation, shell injection, form redaction,
learnings injection, IPv6 SSRF, session permissions, frozen lockfile.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Yunsu <Hybirdss@users.noreply.github.com>
Co-authored-by: Gus <garagon@users.noreply.github.com>
Co-authored-by: Toby Morning <urbantech@users.noreply.github.com>
Co-authored-by: Alberto Martinez <halbert04@users.noreply.github.com>
Co-authored-by: Maksim Soltan <Gonzih@users.noreply.github.com>
Co-authored-by: Ziad Al Sharif <Ziadstr@users.noreply.github.com>
Co-authored-by: Osman Mehmood <mehmoodosman@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-13 07:49:37 -10:00
Garry Tan 115d81d792 fix: security wave 1 — 14 fixes for audit #783 (v0.15.7.0) (#810)
* fix: DNS rebinding protection checks AAAA (IPv6) records too

Cherry-pick PR #744 by @Gonzih. Closes the IPv6-only DNS rebinding gap
by checking both A and AAAA records independently.

Co-Authored-By: Gonzih <gonzih@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: validateOutputPath symlink bypass — resolve real path before safe-dir check

Cherry-pick PR #745 by @Gonzih. Adds a second pass using fs.realpathSync()
to resolve symlinks after lexical path validation.

Co-Authored-By: Gonzih <gonzih@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: validate saved URLs before navigation in restoreState

Cherry-pick PR #751 by @Gonzih. Prevents navigation to cloud metadata
endpoints or file:// URIs embedded in user-writable state files.

Co-Authored-By: Gonzih <gonzih@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: telemetry-ingest uses anon key instead of service role key

Cherry-pick PR #750 by @Gonzih. The service role key bypasses RLS and
grants unrestricted database access — anon key + RLS is the right model
for a public telemetry endpoint.

Co-Authored-By: Gonzih <gonzih@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: killAgent() actually kills the sidebar claude subprocess

Cherry-pick PR #743 by @Gonzih. Implements cross-process kill signaling
via kill-file + polling pattern, tracks active processes per-tab.

Co-Authored-By: Gonzih <gonzih@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix(design): bind server to localhost and validate reload paths

Cherry-pick PR #803 by @garagon. Adds hostname: '127.0.0.1' to Bun.serve()
and validates /api/reload paths are within cwd() or tmpdir(). Closes C1+C2
from security audit #783.

Co-Authored-By: garagon <garagon@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: add auth gate to /inspector/events SSE endpoint (C3)

The /inspector/events endpoint had no authentication, unlike /activity/stream
which validates tokens. Now requires the same Bearer header or ?token= query
param check. Closes C3 from security audit #783.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: sanitize design feedback with trust boundary markers (C4+H5)

Wrap user feedback in <user-feedback> XML markers with tag escaping to
prevent prompt injection via malicious feedback text. Cap accumulated
feedback to last 5 iterations to limit incremental poisoning.
Closes C4 and H5 from security audit #783.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: harden file/directory permissions to owner-only (C5+H9+M9+M10)

Add mode 0o700 to all mkdirSync calls for state/session directories.
Add mode 0o600 to all writeFileSync calls for session.json, chat.jsonl,
and log files. Add umask 077 to setup script. Prevents auth tokens, chat
history, and browser logs from being world-readable on multi-user systems.
Closes C5, H9, M9, M10 from security audit #783.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: TOCTOU race in setup symlink creation (C6)

Remove the existence check before mkdir -p (it's idempotent) and validate
the target isn't already a symlink before creating the link. Prevents a
local attacker from racing between the check and mkdir to redirect
SKILL.md writes. Closes C6 from security audit #783.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: remove CORS wildcard, restrict to localhost (H1)

Replace Access-Control-Allow-Origin: * with http://127.0.0.1 on sidebar
tab/chat endpoints. The Chrome extension uses manifest host_permissions
to bypass CORS entirely, so this only blocks malicious websites from
making cross-origin requests. Closes H1 from security audit #783.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: make cookie picker auth mandatory (H2)

Remove the conditional if(authToken) guard that skipped auth when
authToken was undefined. Now all cookie picker data/action routes
reject unauthenticated requests. Closes H2 from security audit #783.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: gate /health token on chrome-extension Origin header

Only return the auth token in /health response when the request Origin
starts with chrome-extension://. The Chrome extension always sends this
origin via manifest host_permissions. Regular HTTP requests (including
tunneled ones from ngrok/SSH) won't get the token. The extension also
has a fallback path through background.js that reads the token from the
state file directly.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* test: update server-auth test for chrome-extension Origin gating

The test previously checked for 'localhost-only' comment. Now checks for
'chrome-extension://' since the token is gated on Origin header.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: bump version and changelog (v0.15.7.0)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Gonzih <gonzih@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: garagon <garagon@users.noreply.github.com>
2026-04-04 22:12:04 -07:00
Garry Tan 41141007c1 feat: TODOS-aware skills, 2-tier Greptile replies, gitignore fix (#61)
* fix: log non-ENOENT errors in ensureStateDir() instead of silently swallowing

Replace bare catch {} with ENOENT-only silence. Non-ENOENT errors (EACCES,
ENOSPC) are now logged to .gstack/browse-server.log. Includes test for
permission-denied scenario with chmod 444.

* feat: merge TODO.md + TODOS.md into unified backlog with shared format reference

Merge TODO.md (roadmap) and TODOS.md (near-term) into one file organized by
skill/component with P0-P4 priority ordering and Completed section. Add shared
review/TODOS-format.md for canonical format. Add static validation tests.

* feat: add 2-tier Greptile reply system with escalation detection

Add reply templates (Tier 1 friendly, Tier 2 firm), explicit escalation
detection algorithm, and severity re-ranking guidance to greptile-triage.md.

* feat: cross-skill TODOS awareness + Greptile template refs in all skills

/ship Step 5.5: auto-detect completed TODOs, offer reorganization.
/review Step 5.5: cross-reference PR against open TODOs.
/plan-ceo-review, /plan-eng-review: TODOS context in planning.
/retro: Backlog Health metric. /qa: bug TODO context in diff-aware mode.
All Greptile-aware skills now reference reply templates and escalation detection.

* chore: bump version and changelog (v0.3.8)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* docs: update CONTRIBUTING.md for v0.3.8 changes

Clarify test tier cost table (Tier 3 standalone vs combined), add TODOS.md
to "Things to know", mention Greptile triage in ship workflow description.

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-14 20:15:11 -07:00
Garry Tan ff5cbbbfef feat: add remote slug helper and auto-gitignore for .gstack/
- getRemoteSlug() in config.ts: parses git remote origin → owner-repo format
- browse/bin/remote-slug: shell helper for SKILL.md use (BSD sed compatible)
- ensureStateDir() now appends .gstack/ to project .gitignore if not present
- setup creates ~/.gstack/projects/ global state directory
- 7 new tests: 4 gitignore behavior + 3 remote slug parsing
2026-03-14 00:10:00 -05:00
Garry Tan 07b4e15b34 feat: v0.3.2 — project-local state, diff-aware QA, Greptile integration (#36)
* fix: cookie import picker returns JSON instead of HTML

jsonResponse() was defined at module scope but referenced `url` which
only existed as a parameter of handleCookiePickerRoute(). Every API call
crashed, the catch block also crashed, and Bun returned a default HTML
page that the frontend couldn't parse as JSON.

Thread port via corsOrigin() helper and options objects. Add route-level
tests to prevent this class of bug from shipping again.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: add help command to browse server

Agents that don't have SKILL.md loaded (or misread flags) had no way to
self-discover the CLI. The help command returns a formatted reference of
all commands and snapshot flags.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: version-aware find-browse with META signal protocol

Agents in other workspaces found stale browse binaries that were missing
newer flags. find-browse now compares the local binary's git SHA against
origin/main via git ls-remote (4hr cache), and emits META:UPDATE_AVAILABLE
when behind. SKILL.md setup checks parse META signals and prompt the user
to update.

- New compiled binary: browse/dist/find-browse (TypeScript, testable)
- Bash shim at browse/bin/find-browse delegates to compiled binary
- .version file written at build time with git commit SHA
- Build script compiles both browse and find-browse binaries
- Graceful degradation: offline, missing .version, corrupt cache all skip check

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* chore: clean up .bun-build temp files after compile

bun build --compile leaves ~58MB temp files in the working directory.
Add rm -f .*.bun-build to the build script to clean up after each build.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: make help command reachable by removing it from META_COMMANDS

help was in META_COMMANDS, so it dispatched to handleMetaCommand() which
threw "Unknown meta command: help". Removing it from the set lets the
dedicated else-if handler in handleCommand() execute correctly.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* chore: bump version and changelog (v0.3.2)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: add shared Greptile comment triage reference doc

Shared reference for fetching, filtering, and classifying Greptile
review comments on GitHub PRs. Used by both /review and /ship skills.
Includes parallel API fetching, suppressions check, classification
logic, reply APIs, and history file writes.

* feat: make /review and /ship Greptile-aware

/review: Step 2.5 fetches and classifies Greptile comments, Step 5
resolves them with AskUserQuestion for valid issues and false positives.

/ship: Step 3.75 triages Greptile comments between pre-landing review
and version bump. Adds Greptile Review section to PR body in Step 8.
Re-runs tests if any Greptile fixes are applied.

* feat: add Greptile batting average to /retro

Reads ~/.gstack/greptile-history.md, computes signal ratio
(valid catches vs false positives), includes in metrics table,
JSON snapshot, and Code Quality Signals narrative.

* docs: add Greptile integration section to README

Personal endorsement, two-layer review narrative, full UX walkthrough
transcript, skills table updates. Add Greptile training feedback loop
to TODO.md future ideas.

* feat: add local dev mode for testing skills from within the repo

bin/dev-setup creates .claude/skills/gstack symlink to the working tree
so Claude Code discovers skills locally. bin/dev-teardown cleans up.
DEVELOPING_GSTACK.md documents the workflow.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: narrow gitignore to .claude/skills/ instead of all .claude/

Avoids ignoring legitimate Claude Code config like settings.json or CLAUDE.md.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* docs: rename DEVELOPING_GSTACK.md to CONTRIBUTING.md

Rewritten as a contributor-friendly guide instead of a dry plan doc.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* docs: explain why dev-setup is needed in CONTRIBUTING.md quick start

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: add browser interaction guidance to CLAUDE.md

Prevents Claude from using mcp__claude-in-chrome__* tools instead of /browse.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: add shared config module for project-local browse state

Centralizes path resolution (git root detection, state dir, log paths) into
config.ts. Both cli.ts and server.ts import from it, eliminating duplicated
PORT_OFFSET/BROWSE_PORT/STATE_FILE logic.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: rewrite port selection to use random ports

Replace CONDUCTOR_PORT magic offset and 9400-9409 scan with random port
10000-60000. Atomic state file writes, log paths from config module,
binaryVersion field for auto-restart on update.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: move browse state from /tmp to project-local .gstack/

CLI now uses config module for state paths, passes BROWSE_STATE_FILE to
spawned server. Adds version mismatch auto-restart, legacy /tmp cleanup
with PID verification, and removes stale global install fallback.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: update crash log path reference to .gstack/

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* test: add config tests and update CLI lifecycle test

14 new tests for config resolution, ensureStateDir, readVersionHash,
resolveServerScript, and version mismatch detection. Remove obsolete
CONDUCTOR_PORT/BROWSE_PORT filtering from commands.test.ts.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* docs: update BROWSER.md and TODO.md for project-local state

Replace /tmp paths with .gstack/, remove CONDUCTOR_PORT docs, document
random port selection and per-project isolation. Add server bundling TODO.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* docs: update README, CHANGELOG, and CONTRIBUTING for v0.3.2

- README: replace Conductor-aware language with project-local isolation,
  add Greptile setup note
- CHANGELOG: comprehensive v0.3.2 entry with all state management changes
- CONTRIBUTING: add instructions for testing branches in other repos

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat: add diff-aware mode to /qa — auto-tests affected pages from branch diff

When on a feature branch, /qa now reads git diff main, identifies affected
pages/routes from changed files, and tests them automatically. No URL required.
The most natural flow: write code, /ship, /qa.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* chore: update CHANGELOG for complete v0.3.2 coverage

Add missing entries: diff-aware QA mode, Greptile integration,
local dev mode, crash log path fix, README/SKILL.md updates.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 18:10:56 -07:00