Pantheon-Security/medusa

AI-first security scanner. NEW in v2026.7: Claude Code compromise detection — vet .claude/ hooks, permissions & skills before you clone — plus an always-on AI attack-signature scanner and native Rust & PHP rules. Also: medusa scan --git to vet any repo, medusa secrets scan for leaked API keys. 40,000+ patterns, zero setup.

What it solves

MEDUSA is an AI-first security scanner designed to protect AI/ML applications, LLM agents, and RAG pipelines from modern vulnerabilities. It specifically targets AI supply chain attacks, such as repo poisoning and prompt injection, as well as the leakage of sensitive API keys within AI chat histories and shell logs.

How it works

MEDUSA uses over 40,000 detection patterns to scan codebases, configuration files, and histories. It supports a wide range of AI-specific files (like .cursorrules or mcp.json) and traditional code. It can scan local directories or remote GitHub repositories. Additionally, it includes a dedicated secrets scanner that identifies leaked credentials from 21 different issuers (e.g., OpenAI, Anthropic, AWS) and provides an interactive tool to redact them from disk.

Who it’s for

Security researchers, AI developers, and DevOps engineers who need to vet AI-powered tools, agents, and the supply chain of AI-integrated projects.

Highlights

  • AI-Specific Detection: Over 40,000 patterns covering prompt injection, MCP server poisoning, and RAG security.
  • Secrets Purging: Scans AI chat histories (Claude, Cursor, Copilot) and shell histories for leaked API keys with interactive redaction.
  • Repo Poisoning Protection: Detects weaponized AI editor configurations across 28+ file types.
  • Broad Language Support: Supports 79 scanner types across backend, frontend, mobile, and infrastructure languages.
  • Zero Setup: Works immediately after installation via pip without requiring external toolchains for its core rules.

Related

  • Project
    asamassekou10/ship-safe

    An independent security agent for AI-written software that finds, investigates, and verifies vulnerabilities in code, LLM prompts, and agent configurations.

  • Project
    cisco-ai-defense/skill-scanner

    A security scanner for AI Agent Skills that detects prompt injection, data exfiltration, and malicious code using static analysis, behavioral dataflow, and LLM-based semantic analysis.

  • Project
    NVIDIA/SkillSpector

    SkillSpector is an open‑source NVIDIA tool that scans AI‑agent skills (plugins, prompts, code) for security problems. It supports static analysis of 71 known vulnerability patterns, optional LLM‑based semantic checks, live CVE look‑ups, baseline suppression, and multiple output formats (terminal, JSON, Markdown, SARIF). It can run as a CLI, Docker container, or MCP server, making it useful for developers, CI pipelines, and production agents.

  • Project
    vercel-labs/deepsec

    An agent-powered vulnerability scanner for large-scale repositories that uses AI to find and verify deep-seated security issues.

  • Project
    tophant-ai/ClawVault

    A security vault and proxy gateway that protects AI agents by detecting sensitive data leaks, blocking prompt injections, and intercepting dangerous system commands.