AI Security

Security Scanning

Unified security assessment for regulated industries. Modes: static (SAST with semgrep), deps (pip-audit/npm audit), secrets (gitleaks), sbom (CycloneDX). Zero tolerance for medium+ findings. Each finding includes severity, location, fix suggestion, and CWE reference.

When to Use

• Security review, pre-release gate, dependency audit, compliance reporting.
• NOT for code quality metrics -- use /ai-quality.
• NOT for governance compliance -- use /ai-governance.

Modes

static -- SAST

1. Detect stacks -- read project files for active languages.
2. Secret detection -- gitleaks detect --source . --no-git. Any finding is critical.
3. Semgrep -- semgrep scan --config auto --json. Parse for rule IDs, severity, CWE.
4. Manual analysis -- review what tools miss:
   • Authentication on every endpoint (A01)
   • Parameterized queries only (A03)
   • Secrets from env/vault, never hardcoded (A02)
   • HTTP security headers (A05)
   • No user-controlled URLs in HTTP clients (A10)
5. Classify -- severity + OWASP category per finding.

deps -- Dependency Audit

1. Detect lock files -- uv.lock, package-lock.json, Cargo.lock, *.csproj.
2. Run audit -- Python: pip-audit --strict --desc. Node: npm audit --json. Rust: cargo audit --json.
3. Assess exploitability -- mark unreachable paths as reduced severity with justification.
4. Report with upgrade paths.

secrets -- Secret Detection

1. Full scan -- gitleaks detect --source . --no-git --report-format json.
2. Staged scan -- gitleaks protect --staged --no-banner.
3. For each finding: file, line, rule, remediation (rotate credential, store in vault).

sbom -- Software Bill of Materials

1. Generate -- cdxgen -o sbom.json --spec-version 1.5 (CycloneDX JSON).
2. Validate -- all direct deps with versions, license info, package URLs.
3. Flag license risks -- copyleft (GPL, AGPL) conflicting with project license.

--fix -- Auto-fix

When --fix is passed, attempt automatic remediation:

• Secrets: remove from source, add to .gitignore, warn to rotate.
• Dependencies: pip install --upgrade <pkg> for fixable vulns.
• Lint findings: semgrep --autofix where rules support it.
• Report what was fixed and what requires manual intervention.

Severity Classification

Output Contract

# Security Report: [mode]

## Score: N/100
## Verdict: PASS (>=80) | WARN (60-79) | FAIL (<60)

## Findings
| # | Severity | OWASP | CWE | Description | Location | Fix |
|---|----------|-------|-----|-------------|----------|-----|

## Tool Outputs
- gitleaks: [N findings / clean]
- semgrep: [N findings / clean]
- pip-audit: [N findings / clean]

Quick Reference

/ai-security              # run all modes
/ai-security static       # SAST only
/ai-security deps         # dependency audit only
/ai-security secrets      # secret detection only
/ai-security sbom         # generate SBOM
/ai-security deps --fix   # audit + auto-fix

Common Mistakes

• Suppressing findings with # nosec -- fix the root cause or use risk acceptance.
• Ignoring transitive dependency vulns -- they are still exploitable.
• Running gitleaks detect on the full repo for pre-commit -- use gitleaks protect --staged.

Integration

• Pre-commit hook runs gitleaks protect --staged automatically.
• Pre-push hook runs semgrep and pip-audit.
• Release gate (/ai-release) aggregates security results.
• Risk acceptances go to state/decision-store.json via /ai-governance risk.

References

• .ai-engineering/contexts/frameworks/ -- security and OWASP control mapping.
• .ai-engineering/manifest.yml -- non-negotiables and gate thresholds. $ARGUMENTS