security-review（安全审查）

Security Review Skill

Trigger

Run this skill BEFORE installing ANY external package, tool, CLI, npm module, Python library, browser extension, or third-party integration.

No exceptions. "Open source" is not a security clearance.

What This Skill Does

Spawns a security review sub-agent that performs a world-class, current-intelligence security assessment and produces a GO / NO-GO / CONDITIONAL verdict.

How to Use

1. Spawn the review agent

sessions_spawn with model: anthropic/claude-sonnet-4-6, task: [security review prompt below]

2. Standard Review Prompt Template

Fill in [PACKAGE NAME], [INSTALL COMMAND], [DESCRIPTION], [SOURCE URL]:

You are a world-class security analyst. Perform a comprehensive security review before installation.

PACKAGE: [PACKAGE NAME]
Source: [GitHub URL or npm/pypi link]
Install: [INSTALL COMMAND]
Description: [what it claims to do]

Cover ALL 7 sections:

1. LEGITIMACY & TRUST SIGNALS
   - Author/maintainer: GitHub profile, history, reputation
   - Stars, forks, contributors, last commit
   - Red flags: new account, copied code, suspicious activity

2. LATEST SECURITY INTELLIGENCE (USE web_search)
   Search: "[name] security vulnerability", "[name] malware", "[name] CVE",
   "[name] data exfiltration", "[author] security issues", HackerNews/Reddit discussions

3. SOURCE CODE ANALYSIS (USE web_fetch on raw GitHub files)
   - What does the entry point actually do?
   - Network calls — to where?
   - File system access beyond documented scope?
   - Obfuscated code?
   - npm/pip dependencies — any known-bad?

4. DATA FLOW ANALYSIS
   - What data does it access? (conversations, files, env vars, API keys)
   - Where does data go? Local only or external?
   - Telemetry/analytics present?
   - Exfiltration risk for workspace content?

5. PERMISSION SCOPE
   - System access required
   - Network access?
   - Touches env vars or config files?
   - Can it interfere with other processes?

6. DEPENDENCY RISK
   - Full dependency list
   - Any known-vulnerable deps?
   - Recent supply chain attacks on dependencies?

7. VERDICT
   GO / NO-GO / CONDITIONAL
   - Confidence: High/Medium/Low
   - Top 3 specific risks
   - If CONDITIONAL: exact conditions required before install

Write complete review to:
C:\Users\hhx-sandbox2\.openclaw\workspace\logs\security-review-[package-name]-[YYYY-MM-DD].md

Verify file exists and is complete before reporting done.

3. Read the output

After the sub-agent completes, read the file and surface the verdict to Marcus.

4. Marcus decides

Present the verdict clearly. Never install without explicit Marcus approval after a GREEN or CONDITIONAL review.

Standing Rules

• NO install without a completed security review — not even "quick" ones
• Web search is mandatory — latest CVEs and community reports must be checked
• Source code must be read — not just the README
• All reviews logged to logs/security-review-[package]-[date].md
• MEMORY.md updated with verdict after each review

Review Log Index