c11244778d
The first round of this PR removed SKIP_WIN32, fixed venv_py to use Scripts/python.exe, and added Lib/site-packages to the consumer glob — all necessary. Windows verification (Win11 ARM64, Py 3.13, Git Bash) showed two more blockers, both addressed here. 1. Pip dependency resolver picks unbuildable cryptography on ARM64. Without --prefer-binary, pip picks a cryptography version with no published ARM64 wheel and tries to build it from source. That needs Rust/Cargo, almost never present on user machines → BUILD_FAILED with err_kind=other:cryptography. A binary wheel exists for an adjacent version (cryptography-46.0.3-cp311-abi3-win_arm64.whl); --prefer-binary tells pip to pick it. Cross-platform safe (no-op where the latest version already has a wheel). 2. pywin32 .pth files aren't processed by sys.path.insert(). With the venv built, ensure_agent_sdk.py's post-build probe passes (it runs from venv_py, where Python's site.py at startup processes pywin32.pth and registers win32/, win32/lib/ plus runs pywin32_bootstrap.py to set the DLL search dir). But llm.py runs in the hook's SYSTEM Python and adds the venv via sys.path.insert(), which doesn't trigger site.py at all. Without the bootstrap, the SDK's mcp.client.stdio → mcp.os.win32.utilities chain raises ModuleNotFoundError: pywintypes and the agentic reviewer falls back to single-shot silently — exactly the symptom this PR is trying to fix. The probe says NOOP_VENV; the actual consumer fails. Probe and consumer use different Pythons. Replicate what site.py would do: after inserting site-packages, also insert win32/ and win32/lib/, then exec pywin32_bootstrap.py. Pulled into a shared helper _inject_agent_sdk_venv_into_syspath() so both consumer sites (3P SDK fallback, agentic_review fallback) call the same code — Windows handling stays in one place. Verified on macOS (POSIX path unchanged): - Helper end-to-end test: POSIX-layout venv detected + fake package imports successfully via the injected path - Windows-layout venv also detected; win32 branch correctly skipped via sys.platform check - Both files pass py_compile Credit: @mhegazy verified the previous commit on Win11 ARM64 / Py 3.13 / Git Bash, surfaced both issues end-to-end, and provided the exact fix patterns. This commit applies them with the pywin32 part factored into a shared helper (vs. inlining at both consumer sites). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
security-guidance
Security review for Claude-generated code. Three layers:
- Pattern warnings — instant regex-based reminders on
Edit/Writefor ~25 known-dangerous patterns (yaml.load,torch.load(weights_only=False),pickle.loadon untrusted data, rawinnerHTML, hardcoded secrets, etc.). - LLM diff review — when Claude finishes a turn, the plugin sends the diff to a fast LLM call (Opus 4.7 by default) and feeds high-severity findings back to Claude so it can fix them before you see the response.
- Agentic commit review — on
git commit, an SDK-driven reviewer reads related files (Read/Grep/Glob) to trace data flow across the codebase, catching multi-file vulnerabilities pattern matching misses (IDOR, auth bypass, cross-file SSRF).
Findings cover common web-vulnerability classes — injection, XSS, SSRF, hardcoded secrets, IDOR, auth bypass, unsafe deserialization, and path traversal among others.
Install
/plugin install security-guidance@claude-plugins-official
Marketplace ships enabled by default in Claude Code — no setup beyond having the CLI itself.
Prerequisites
- Claude Code CLI ≥ v2.1.144
- Python 3.8+ on
PATH(python3,python, orpy -3— the plugin picks the first that works) - A working API path (subscription, API key, or 3P provider config)
Configuration
All configuration is via environment variables. None are required for default behavior.
Selecting a model
# 1P / gateway: a canonical model id
SECURITY_REVIEW_MODEL=claude-opus-4-7 # default
# Bedrock: use the inference-profile id
SECURITY_REVIEW_MODEL=us.anthropic.claude-opus-4-7
# Vertex: use the Vertex date-tag form
SECURITY_REVIEW_MODEL=claude-opus-4-7@20260218
SECURITY_REVIEW_MODEL controls the LLM diff review. SG_AGENTIC_MODEL (same syntax) controls the agentic commit reviewer; defaults to the same model.
Enabling/disabling layers
| Variable | Default | What it does |
|---|---|---|
SECURITY_GUIDANCE_DISABLE=1 |
unset | Kill switch — disables the entire plugin |
ENABLE_PATTERN_RULES=0 |
on | Disable layer 1 (regex pattern warnings) |
ENABLE_CODE_SECURITY_REVIEW=0 |
on | Disable all LLM reviews (Stop hook + commit/push) |
ENABLE_STOP_REVIEW=0 |
on | Disable only the Stop-hook diff review, keeping commit/push reviews. Useful for multi-agent / shared-worktree setups where another agent can move HEAD between a worker's turns |
ENABLE_COMMIT_REVIEW=0 |
on | Disable layer 3 (agentic commit review) |
Higher-recall mode
SG_DUAL_OR=on # default off
Runs two parallel review calls and unions the findings. Catches a few percentage points more vulnerabilities in our testing, at roughly 2× the API cost per review. Most users don't need it.
Org-specific policies
Drop a claude-security-guidance.md in any of:
~/.claude/claude-security-guidance.md— user-wide rules<project>/.claude/claude-security-guidance.md— project rules, intended to be committed<project>/.claude/claude-security-guidance.local.md— local overrides, intended to be.gitignore'd
All three are loaded and concatenated into the LLM diff review's prompt in the order user → project → project-local. If the combined size exceeds the 8 KB prompt budget, the tail is truncated, so user-wide rules are kept and project-local rules are dropped first. The agentic commit reviewer (layer 3) does not currently read this file. Example:
# Acme security rules
- All SELECTs against the `customers` or `orders` tables MUST go through `db.replica`,
never `db.primary`. Primary is for writes only.
- Background jobs must not use the user-context auth token; they get
service-account creds from `jobs.get_service_account()`.
- Calls to `requests.get(url)` with a user-controlled `url` need
the SSRF-allowlist wrapper at `acme.net.safe_request`.
Built-in rules cover common web-vulnerability classes without it — claude-security-guidance.md is for things specific to your codebase that the model can't infer.
Privacy and data handling
The plugin sends data to a model endpoint to perform its reviews. Specifically, each Stop-hook diff review transmits the changed file paths, the diff hunks, and the relevant file contents in the diff; each agentic commit review additionally transmits any files the reviewer pulls in via Read/Grep/Glob while tracing data flow. Your claude-security-guidance.md contents (user, project, and local) are appended to the prompt on every review, so don't put secrets in it.
Where that data goes depends on your Claude Code configuration:
- Default (Anthropic API / subscription): sent to
api.anthropic.comand handled under Anthropic's Commercial Terms and Privacy Policy. - LLM gateway (
ANTHROPIC_BASE_URLset): sent to your gateway URL instead. The gateway operator's terms apply. - 3rd-party providers (Bedrock / Vertex / Foundry / Mantle): sent to your configured provider endpoint. The provider's data-handling terms apply (e.g., AWS / GCP / Azure).
The plugin writes its own debug log to ~/.claude/security/log.txt (override with SECURITY_GUIDANCE_DEBUG_LOG). The log contains diffstate metadata and finding categories — no full file contents or model prompts — and rotates at 1 MB. Nothing is uploaded.
Limitations
This is a best-effort assistive tool, not a guarantee. Treat findings as suggestions, not as a substitute for human code review, SAST/DAST, dependency scanning, or pen-testing. The reviewer can miss vulnerabilities, produce false positives, and may behave differently across codebases, languages, and model versions. No warranty is provided — use is subject to Anthropic's Commercial Terms.
Troubleshooting
Plugin doesn't seem to fire — check that ~/.claude/claude-security-guidance.md (or hook activity) shows in debug logs. Run Claude Code with --debug-file /tmp/claude/debug.txt and grep for security_reminder_hook. The plugin also writes its own log to ~/.claude/security/log.txt.
Review never finds anything — verify your API path works. On 3P providers, check SECURITY_REVIEW_MODEL is set to a provider-specific id (not a bare claude-opus-4-7). On LLM gateways, check the gateway's logs for POST /v1/messages traffic from the plugin.
Too many false positives — drop SECURITY_REVIEW_MODEL to a cheaper model (claude-sonnet-4-6) and re-evaluate; if precision is the priority, stay on Opus 4.7.
Want to silence a specific finding — add a comment to the line explaining why it's safe; the LLM reviewer treats inline justifications as exclusions. For systemic exclusions, document them in your claude-security-guidance.md.
Reporting issues
Open an issue on the security-guidance plugin repo with:
- The Claude Code CLI version (
claude --version) - Provider setup (1P / Bedrock / Vertex / LLM gateway / etc.)
- A minimal repro diff
- The relevant section of
~/.claude/security/log.txt