skills.rest
rinchsan
rinchsanCOMMUNITY
Explore all skills by this author →
𝕏@rinchsanGitHub

security-scan

Scan Claude Code configuration files for security vulnerabilities and misconfigurations using AgentShield.

5|Updated Jul 8, 2019
Search Tags:#claude code#mcp#security scan#prompt injection#configuration audit#ci cd#agentshield
One-click install
npx skills add https://github.com/rinchsan/dotfiles --skill security-scan-rinchsan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: security-scan
Source: https://github.com/rinchsan/dotfiles/tree/main/.claude/skills/security-scan
Command: npx skills add https://github.com/rinchsan/dotfiles --skill security-scan-rinchsan

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ecc-agentshield.

What problem does it solve? Claude Code configurations like CLAUDE.md, settings.json, and MCP server definitions can contain hardcoded secrets, overly permissive permissions, and prompt injection risks that are easy to miss during manual review. This Skill audits the entire .claude/ directory and reports graded, actionable findings. ## Core Features & Use Cases - Configuration Auditing: Scans CLAUDE.md, settings.json, mcp.json, hooks, and agent definitions for secrets, dangerous permissions, and injection patterns. - Auto-Fix and CI Integration: Applies safe automatic fixes and exports JSON, Markdown, or HTML reports suitable for GitHub Actions pipelines. - Deep Adversarial Analysis: Runs an optional three-agent red team, blue team, and auditor pipeline for deeper threat assessment. - Use Case: Before committing changes to .claude/settings.json, run a scan to catch a wildcard Bash(*) permission or a hardcoded API key before it reaches production. ## Quick Start Ask the assistant to run a security scan on the current project's .claude directory and summarize any critical or high severity findings.

Frequently Asked Questions about security-scan

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I scan my Claude Code configuration for security issues?▼

Run npx ecc-agentshield scan in your project root to audit the .claude directory. The scan checks CLAUDE.md, settings.json, mcp.json, hooks, and agent files, then outputs a graded report with severity levels.

What security problems does AgentShield detect in Claude Code configs?▼

It detects hardcoded API keys and tokens, wildcard Bash permissions, command injection in hooks via interpolation, risky MCP servers, auto-run instructions in CLAUDE.md, and missing deny lists. Findings are graded from A to F by severity.

Can I run a Claude Code security scan in CI/CD?▼

Yes, use the affaan-m/agentshield GitHub Action with a path, minimum severity, and fail-on-findings option. You can also run npx ecc-agentshield scan --format json to integrate results into any pipeline.

Does the security scan automatically fix vulnerabilities?▼

Yes, running with the --fix flag applies safe automatic fixes such as replacing hardcoded secrets with environment variable references and tightening wildcard permissions. Suggestions marked manual-only are never modified.

Why does the deep analysis mode require an Anthropic API key?▼

The --opus flag runs a three-agent adversarial pipeline with attacker, defender, and auditor roles powered by Claude models. This requires the ANTHROPIC_API_KEY environment variable to authenticate API calls.

Related Skills

Explore More Software Engineering Skills→
emiled16emiled16

security-scan

Audit Claude Code configuration files for security vulnerabilities and misconfigurations.

Community
Intermediate
0xHoneyJar0xHoneyJar

audit-env

Scan environment files, client code, and git history for secret leakage.

Official
Advanced
FlorianBruniauxFlorianBruniaux
5.9k

security-audit

Audits project code and Claude Code configuration for secrets, injection vectors, and vulnerabilities with a scored report.

Community
Advanced
voidrotvoidrot

security-scan

Scan Claude Code and agent configurations for security vulnerabilities.

Community
Advanced