MCP & Claude Code security scanner — threat-models plugins, MCP servers, hooks, skills & connectors with an LLM before you trust them. Catches prompt injection, tool poisoning & credential exfiltratio
proluct/skannaSecurity scanner for Claude Code skills, plugins, and MCP servers. A verdict before you install.
davidnichols-ops/trustcardCryptographic trust infrastructure for MCP servers — signed manifests, TOFU pinning, two-gate enforcement, and the "npm audit" scanner.
eltociear/mcp-auditScan MCP servers & AI-agent skills for malicious patterns (prompt injection in tool descriptions, credential exfil, download-and-execute). 17 patterns / 60 signatures, zero-dep. Measured over 196 publ