🔍 ISNAD Scan Action

Security scanner for AI agent skills. Detect dangerous patterns before they reach production.

Features

• 🛡️ 69+ detection patterns — shell injection, credential theft, filesystem access, network exfiltration
• ⚡ Fast — scans in seconds, not minutes
• 🔧 Configurable — fail on WARN, DANGER, or never
• 📊 Rich output — detailed findings with line numbers

Quick Start

name: Security Scan

on: [push, pull_request]

jobs:
  scan:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: counterspec/isnad-scan-action@v1
        with:
          path: './skills'
          fail-on: 'DANGER'

Inputs

Input	Description	Default
path	Path to scan	.
fail-on	Fail on trust level (WARN, DANGER, never)	DANGER
verbose	Show verbose output	false

Outputs

Output	Description
trust-level	Result: SAFE, WARN, or DANGER
findings-count	Number of security findings
report	Full scan report

Example: Comment on PR

- uses: counterspec/isnad-scan-action@v1
  id: scan
  with:
    path: './skills'
    fail-on: 'never'  # Don't fail, just report

- name: Comment on PR
  if: github.event_name == 'pull_request'
  uses: actions/github-script@v7
  with:
    script: |
      const trustLevel = '${{ steps.scan.outputs.trust-level }}';
      const emoji = trustLevel === 'SAFE' ? '✅' : trustLevel === 'WARN' ? '⚠️' : '🚨';
      
      github.rest.issues.createComment({
        owner: context.repo.owner,
        repo: context.repo.repo,
        issue_number: context.issue.number,
        body: `## ${emoji} ISNAD Security Scan\n\n**Trust Level:** ${trustLevel}\n**Findings:** ${{ steps.scan.outputs.findings-count }}\n\n\`\`\`\n${{ steps.scan.outputs.report }}\n\`\`\``
      });

Trust Levels

Level	Description
✅ SAFE	No concerning patterns detected
⚠️ WARN	Patterns that need review (env access, file writes)
🚨 DANGER	High-risk patterns (shell exec, credential access, network calls)

Links

• ISNAD Protocol
• isnad-scan on PyPI
• Report Issues

---

Built by ISNAD Protocol — Trust infrastructure for AI agents.