Skip to content

v0.12.0 — MCP tool-argument taint (MCP013/MCP014) + installed command

Latest
Latest

Choose a tag to compare

View all tags
@akanthed akanthed released this 04 Oct 14:21
· 1 commit to main since this release
v0.12.0
56a560b
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

The scanner now finds command injection and path traversal inside MCP servers, the two largest classes of real MCP CVEs. New rules MCP013 (MCP tool argument reaches a shell command) and MCP014 (tool argument used as a file path without containment) treat tool arguments as the taint source. They were validated against six published MCP command-injection CVEs (detected on the vulnerable release, clean on the patched one) and 25 popular MCP servers (no default-tier findings).

Also new: secureai-scan installed audits the MCP servers and Agent Skills already installed on your machine (Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, Gemini CLI, Cline, Roo Code, Amazon Q), and npm run ecosystem scans a list of public MCP servers for triage. Nine real-world false-positive classes are fixed at the root, and AI001 now catches Fetch-API request handlers (Next.js App Router, Remix, SvelteKit, Hono, Workers) and resolves tsconfig path aliases.

See CHANGELOG.md for the full list.

What's Changed

  • chore: update version references to 0.11.0 in documentation by @akanthed in #34
  • feat: add MCP012/AI013/AI014 detection rules, fix a real MCP011 FP by @akanthed in #35
  • 0.12.0: MCP tool-argument taint (MCP013/MCP014), installed command, nine false-positive fixes by @akanthed in #42

Full Changelog: v0.11.0...v0.12.0

Contributors

akanthed
Assets 2
Loading