Static security audit for Claude Code plugins, skills, agents, hooks, and MCP servers. Detects backdoors, prompt injection, persistence hooks, and supply-chain risks via parallel sub-agents.
Agent
Claude Code artifact auditor sub-agent of the plugin-auditor plugin. Invoked by the audit skill orchestrator to scan SKILL.md, agents, slash commands, hooks, MCP server declarations, settings.json, and CLAUDE.md files for prompt injection, persistence hooks, context exfiltration, history theft, trigger hijacking, and…
Agent
Configuration auditor sub-agent of the plugin-auditor plugin. Invoked by the audit skill orchestrator to scan Claude Code settings.json, CI/CD workflows (GitHub Actions, GitLab CI, CircleCI, Travis, Jenkins), Dockerfile and docker-compose, setup/install/bootstrap scripts, devcontainer configs, and editor configs for…
Agent
Prompt-injection guard sub-agent of the plugin-auditor plugin. A second, independent detector that reads only LLM-steering artifacts (SKILL.md, agents, commands, CLAUDE.md, hooks, .mcp.json, plugin.json descriptions, and any .md with name/description frontmatter) and judges each for injection intent. Runs on a…
Agent
Network and filesystem auditor sub-agent of the plugin-auditor plugin. Invoked by the audit skill orchestrator to extract every URL in the repository, classify each host against an allowlist, then scan for filesystem-scope violations (path traversal, access to /.ssh, /.aws, browser state, Claude Code paths), data…
Agent
Static code auditor sub-agent of the plugin-auditor plugin. Invoked by the audit skill orchestrator to scan a repository for hardcoded credentials, obfuscated payloads, dangerous shell patterns, committed binaries, hidden state files, modifications to global dotfiles, and OS-level persistence. Returns a structured…
Agent
Supply-chain auditor sub-agent of the plugin-auditor plugin. Invoked by the audit skill orchestrator to scan dependency manifests (package.json, requirements.txt, pyproject.toml, Cargo.toml, go.mod, Gemfile, composer.json) and lockfiles for lifecycle scripts, typosquatting heuristics, missing or unfrozen lockfiles…