Skip to content

Add skill-audit-mcp to Testing#134

Open
eltociear wants to merge 1 commit into
devsecops:masterfrom
eltociear:add-skill-audit-mcp
Open

Add skill-audit-mcp to Testing#134
eltociear wants to merge 1 commit into
devsecops:masterfrom
eltociear:add-skill-audit-mcp

Conversation

@eltociear
Copy link
Copy Markdown

Adds skill-audit-mcp under Testing.

Static security scanner specifically for the MCP server and AI agent skill layer — where most production prompt-injection and credential-exfiltration attacks land in 2026 DevSecOps pipelines.

Detects 68 attack patterns across 4 severity levels:

  • CRITICAL — credential exfiltration, seed-phrase harvest, download-and-execute
  • HIGH — arbitrary code execution, auth bypass, identity impersonation
  • MEDIUM — prompt injection (instruction override, hidden Unicode, encoded payloads)
  • LOW — external URL refs, broad filesystem access

Distribution for DevSecOps pipelines:

  • GitHub Action: uses: eltociear/skill-audit-mcp@v1 → SARIF → GitHub Code Scanning
  • Multi-arch Docker: ghcr.io/eltociear/skill-audit-mcp:v1 (amd64+arm64)
  • MCP server transport
  • Hosted x402 API (pay-per-scan, USDC micropayments on Base)

Zero runtime dependencies. Python 3.6+. MIT.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant