Scan Claude Code configuration files for security vulnerabilities and injection risks.
npx skills add https://github.com/gganbukim1/myskills --skill security-scan-gganbukim1
Please help me install this Agent Skill. Skill: security-scan Source: https://github.com/gganbukim1/myskills/tree/main/security-scan Command: npx skills add https://github.com/gganbukim1/myskills --skill security-scan-gganbukim1
It helps you detect and reduce security risks in your Claude Code setup before misconfigurations, prompt injection paths, or unsafe tool access make it into production.
.claude/ content for secrets, risky permissions, and injection patterns that commonly lead to compromise.CLAUDE.md, .claude/settings.json, mcp.json, hooks/, and agents/*.md for unsafe directives, tool access, and command injection surfaces.Example use case: After updating your repository’s Claude configuration for a new MCP server and custom hooks, run this scan to catch hardcoded tokens, overly permissive allow lists, and shell/command injection risks before you commit.
Ask your AI assistant to run an AgentShield scan over your project to produce a security report for the .claude/ directory.
High-intent search queries and answers about installing and using this skill.
To scan Claude Code configuration for vulnerabilities, run an AgentShield scan over your project to audit `.claude/` directories, which checks for secrets, risky permissions, and injection patterns.
Prompt injection risks in MCP server definitions are detected by reviewing `mcp.json` for unsafe directives, risky tool access, and command injection surfaces that could compromise your setup.
Yes, you can automatically fix misconfigurations in `CLAUDE.md` and `.claude/settings.json` using the optional auto-fix feature for safe remediations after the scan produces severity-graded findings.
Yes, the security scan checks hook scripts under `hooks/` for command injection surfaces, alongside reviewing agent configuration files and MCP server definitions for unsafe tool access.
Security findings for Claude Code setups are reported as severity-graded scores from A to F, allowing you to prioritize fixes for hardcoded tokens, overly permissive allow lists, and shell injection risks.
You should run a configuration vulnerability scan on your `agents/*.md` directory after updating repository configurations for new MCP servers and custom hooks, ensuring no unsafe directives reach production.
cescrafliScan .claude directory configurations for security vulnerabilities and misconfigurations.
timequityScan generated code for OWASP vulnerabilities and produce security reports.
mahg-esDesign and enforce CI/CD quality gates for code, tests, security, build, E2E, and performance.
getsentryScan agent skills for security vulnerabilities using static analysis and pattern matching.