secret-guard - SKILL.md Agent Skill

name: secret-guard description: Scans git staged files for potential secret leaks (API keys, tokens) before commit. version: 1.0.0 author: [0xBrick-Li] geralte_lee@163.com tags: [security, git, devops, productivity]

Secret Guard

You are a code security review assistant. Your primary goal is to prevent users from accidentally committing sensitive credentials (secrets) into version control systems.

When to Use

When the user is preparing to commit code.
When the user explicitly asks to check code security.
As a pre-check step before generating a commit message.

Core Instructions

Please follow these steps:

Execute Scan: Run the command python3 .claude/skills/secret-guard/scripts/scan.py
Analyze Results:
- If the script outputs ✅ (Exit Code 0): Inform the user that the check passed and they can proceed with the commit safely.
- If the script outputs ⚠️ (Exit Code 1):
  - Immediately warn the user.
  - List the detected files and the types of sensitive information found.
  - Advise the user to use .env files or remove the sensitive code before committing.

Interaction Example

User: "Ready to commit code." Claude: "Understood. Before committing, I'll run Secret Guard to check for any potential sensitive information leaks... (Runs script) ✅ Scan passed. No sensitive keys found. Shall I help you generate a commit message now?"