supply-chain-audit

Audit the software supply chain for $ARGUMENTS.

Follow every step below. The output must be a structured assessment with specific findings and remediation actions — not a generic checklist.

---

Step 1: Assess Source Integrity

Evaluate how the source code is protected from tampering:

Checks

Control	How to Verify	Severity if Missing
Signed commits	Check git log --show-signature or .gitsign config	MEDIUM
Branch protection	Check for CODEOWNERS, branch protection rules, or ruleset config	HIGH
Code review requirement	Check PR merge rules — are reviews required? How many?	HIGH
CODEOWNERS file	Look for CODEOWNERS or .github/CODEOWNERS	MEDIUM
Force push disabled	Check branch protection settings	HIGH
Commit signing policy	Check for commit.gpgSign in .gitconfig or CI enforcement	LOW

Evidence Collection

# Check for CODEOWNERS
find . -name "CODEOWNERS" -type f

# Check for branch protection config (GitHub)
find . -path "*/.github/*" -name "*.yml" | head -20

# Check for signed commits
git log --format="%H %G?" -10

# Check for merge commit policy
grep -r "merge" .github/ --include="*.yml" -l

Document each control as PASS, FAIL, or PARTIAL with specific evidence.

---

Step 2: Assess Build Integrity

Evaluate whether the build process can be tampered with:

Checks

Control	How to Verify	Severity if Missing
Build runs on hosted infrastructure	CI config uses hosted runners, not self-hosted	HIGH
Build is defined in code	CI/CD config is in the repo, not configured in the UI	HIGH
Build is hermetic	No network access during build (or explicitly declared dependencies)	MEDIUM
Build is reproducible	Same inputs produce same outputs (lockfiles, pinned versions)	MEDIUM
Build provenance generated	Attestation or SLSA provenance document created	HIGH
Build logs retained	Logs are stored and accessible for audit	MEDIUM

Evidence Collection

# Find CI/CD configuration files
find . -name "*.yml" -path "*/.github/workflows/*" -o -name ".gitlab-ci.yml" -o -name "Jenkinsfile" -o -name "azure-pipelines.yml" | head -20

# Check for self-hosted runners
grep -r "self-hosted" .github/workflows/ --include="*.yml"

# Check for build provenance generation
grep -r "slsa\|provenance\|attest\|sigstore\|cosign" .github/workflows/ --include="*.yml"

# Check for hermetic build indicators
grep -r "network.*none\|--network=none\|hermetic" .github/workflows/ --include="*.yml"

---

Step 3: Assess Dependency Integrity

Evaluate how third-party dependencies are managed:

Checks

Control	How to Verify	Severity if Missing
Lockfiles present and committed	package-lock.json, poetry.lock, go.sum, Cargo.lock exist in repo	CRITICAL
Lockfiles are not ignored	Not listed in .gitignore	CRITICAL
Dependencies pinned to exact versions	No floating ranges (^, ~, *) in lockfile	HIGH
Dependency review on PRs	GitHub Dependency Review Action or equivalent	MEDIUM
Private registry configured	.npmrc, pip.conf, or nuget.config points to approved registry	MEDIUM
Typosquat check	Package names verified against known typosquat lists	MEDIUM
Vulnerability scanning	Dependabot, Renovate, Snyk, or npm audit in CI	HIGH
License compliance	License checker in CI or documented policy	LOW

Evidence Collection

# Find lockfiles
find . -maxdepth 3 \( -name "package-lock.json" -o -name "yarn.lock" -o -name "pnpm-lock.yaml" -o -name "poetry.lock" -o -name "Pipfile.lock" -o -name "go.sum" -o -name "Cargo.lock" -o -name "composer.lock" \) -type f

# Check if lockfiles are gitignored
grep -r "lock" .gitignore 2>/dev/null

# Check for registry configuration
find . -maxdepth 2 \( -name ".npmrc" -o -name "pip.conf" -o -name "nuget.config" -o -name ".yarnrc.yml" \) -type f

# Check for dependency scanning in CI
grep -r "dependabot\|renovate\|snyk\|npm audit\|pip-audit\|cargo audit\|trivy" .github/ --include="*.yml" -l 2>/dev/null

# Check for floating versions in package.json
grep -E '"[~^*]' package.json 2>/dev/null | head -10

---

Step 4: Assess Artifact Integrity

Evaluate how built artifacts are protected:

Checks

Control	How to Verify	Severity if Missing
Container images signed	Cosign or Notation signatures	HIGH
SBOMs generated	SPDX or CycloneDX SBOM per release	HIGH
Deployment requires verified signature	Admission controller or policy engine validates signatures	HIGH
Artifact stored in trusted registry	ECR, GCR, ACR, or self-hosted with access controls	MEDIUM
Image tags are immutable	Tags cannot be overwritten (or digests used instead of tags)	MEDIUM
Attestation published	In-toto attestation or SLSA provenance attached to artifact	HIGH

Evidence Collection

# Check for image signing in CI
grep -r "cosign\|notation\|sigstore" .github/workflows/ --include="*.yml" 2>/dev/null

# Check for SBOM generation
grep -r "sbom\|spdx\|cyclonedx\|syft\|trivy.*sbom" .github/workflows/ --include="*.yml" 2>/dev/null

# Check for Dockerfile
find . -maxdepth 3 -name "Dockerfile*" -type f

# Check for registry push configuration
grep -r "docker push\|docker/build-push\|ecr\|gcr\|acr" .github/workflows/ --include="*.yml" 2>/dev/null

---

Step 5: Determine SLSA Level Achieved

Based on the evidence collected, determine the SLSA level:

SLSA Level	Requirements	Key Evidence
Level 0	No guarantees	No build provenance, no lockfiles
Level 1	Documentation of the build process	Build defined in code, lockfiles present
Level 2	Hosted build service, signed provenance	CI runs on hosted runners, provenance generated
Level 3	Hardened build, non-falsifiable provenance	Hermetic build, isolated build environment, tamper-proof provenance
Level 4	Hermetic + reproducible, two-party review	All of Level 3 + reproducible builds + enforced code review

Assessment

For each level, document:

1. Requirements met — specific evidence from Steps 1-4
2. Requirements not met — specific gaps with remediation steps
3. Current level — the highest level where ALL requirements are met

---

Step 6: Produce Findings and Recommendations

Finding Format

For each finding:

Field	Content
ID	SC-[NNN]
Title	Short description
Severity	CRITICAL / HIGH / MEDIUM / LOW
Category	Source / Build / Dependency / Artifact
Current state	What was found (with evidence)
Risk	What could happen if not addressed
Recommendation	Specific action to take
Effort	S / M / L
SLSA impact	Which SLSA level this blocks

Prioritisation

Order findings by:

1. CRITICAL severity first (these block production readiness)
2. Then by SLSA impact (findings that unlock the next SLSA level)
3. Then by effort (quick wins first)

---

Rules

• Lockfiles are non-negotiable — any project without committed lockfiles gets an automatic CRITICAL finding. No exceptions.
• Every CI/CD action must be pinned to a SHA, not a tag — tags can be moved maliciously. actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 not actions/checkout@v4. Reference the GitHub Actions security hardening guide.
• SBOMs should be generated for every release artifact — without an SBOM, you cannot respond to a vulnerability disclosure (you do not know if you are affected).
• Findings must cite specific evidence — "dependency scanning is missing" is not a finding. "No npm audit, pip-audit, Snyk, or Dependabot configuration found in .github/workflows/ or dependabot.yml" is a finding.
• Do not recommend tools without justification — if recommending Cosign for image signing, explain why it fits this project's stack and CI platform.
• Reference SLSA framework for supply chain levels and Sigstore for keyless signing.

---

Output Format

# Supply Chain Security Audit: [Repository / System]

**Date:** [YYYY-MM-DD]
**Auditor:** Claude (automated assessment)
**SLSA Level Achieved:** [0-4]

## Executive Summary

[2-3 sentences: current posture, most critical gaps, recommended priority.]

## Assessment by Area

### Source Integrity
| Control | Status | Evidence |
|---------|--------|----------|
| Signed commits | PASS/FAIL/PARTIAL | [specific evidence] |
| Branch protection | PASS/FAIL/PARTIAL | [specific evidence] |
| ... | ... | ... |

### Build Integrity
| Control | Status | Evidence |
|---------|--------|----------|
| ... | ... | ... |

### Dependency Integrity
| Control | Status | Evidence |
|---------|--------|----------|
| ... | ... | ... |

### Artifact Integrity
| Control | Status | Evidence |
|---------|--------|----------|
| ... | ... | ... |

## SLSA Level Assessment

| Level | Status | Blocking Gaps |
|-------|--------|---------------|
| Level 1 | MET/NOT MET | [gaps] |
| Level 2 | MET/NOT MET | [gaps] |
| Level 3 | MET/NOT MET | [gaps] |
| Level 4 | MET/NOT MET | [gaps] |

## Findings

### SC-001: [Title]
- **Severity:** CRITICAL
- **Category:** Dependency
- **Current state:** [evidence]
- **Risk:** [what could happen]
- **Recommendation:** [specific action]
- **Effort:** S
- **SLSA impact:** Blocks Level 1

### SC-002: [Title]
...

## Remediation Roadmap

| Priority | Finding | Effort | SLSA Impact |
|----------|---------|--------|-------------|
| 1 | SC-001: [title] | S | Blocks L1 |
| 2 | SC-003: [title] | M | Blocks L2 |
| ... | ... | ... | ... |

---

Related Skills

• /security-engineer:dependency-audit — deep-dive into dependency vulnerabilities. Use after this audit identifies dependency gaps.
• /devops:write-pipeline — harden the CI/CD pipeline based on audit findings. Pin actions to SHAs, add SBOM generation, add image signing.