name: cm-secret-shield description: "Defense-in-depth security for AI-assisted development. Pre-commit secret scanning (Gitleaks + native fallback), repo-wide pattern detection, token lifecycle management, and AI agent safety rules. Prevents secret leaks at every stage: write → commit → push → deploy. Use BEFORE any git push or deploy, or when setting up new projects."
🛡️ Secret Shield — Defense-in-Depth Security
Secrets leak at FIVE stages. Guard ALL five. Write → Commit → Push → Deploy → Runtime One missed stage = one leaked key = one compromised system.
The Iron Laws
NEVER commit secrets. EVER.
NEVER output secrets in logs, chat, or AI responses.
NEVER trust .gitignore alone — it doesn't protect git history.
PRE-COMMIT HOOKS are your FIRST line of defense.
ROTATION is not optional after a leak.
When to Use
ALWAYS when:
- Setting up a new project (called by
cm-project-bootstrapPhase 0.5) - Before first
git pushon any project - After discovering a potential secret leak
- Setting up CI/CD pipelines
- Reviewing security posture of existing projects
- User says: "check secrets", "security audit", "leaked key", "rotate token"
Integrates with:
cm-project-bootstrap— Security Foundation phasecm-safe-deploy— Gate 0 enhanced secret hygienecm-test-gate— Layer 5 security scancm-identity-guard— Token lifecycle management
The 5 Defense Layers
┌─────────────────────────────────────────────────────────┐
│ Layer 1: WRITE GUARD — AI agent behavior rules │
│ Layer 2: PRE-COMMIT GUARD — Block secrets at commit │
│ Layer 3: REPO SCAN — Full repo pattern check │
│ Layer 4: DEPLOY GATE — Pre-deploy secret audit │
│ Layer 5: RUNTIME GUARD — Env var hygiene & rotation│
└─────────────────────────────────────────────────────────┘
Layer 1: Write Guard — AI Agent Safety Rules
Goal: Prevent the AI agent itself from introducing secrets into code.
Rules for AI Agents
✅ DO:
- Use environment variables: process.env.SECRET_KEY
- Use .dev.vars for local development
- Use platform-specific secret stores: wrangler secret put, Supabase vault
- Mask secrets in logs: console.log('Key:', key.slice(0,4) + '***')
- Reference secret NAMES, not VALUES
❌ NEVER:
- Hardcode API keys, tokens, or passwords in source code
- Put secrets in wrangler.jsonc, package.json, or any tracked file
- Output full secret values in chat, logs, or error messages
- Use placeholder secrets that look real (e.g., sk-1234567890abcdef)
- Store secrets in i18n files, README, or documentation
Secret Patterns to NEVER Generate
// ❌ NEVER write code like this:
const API_KEY = "sk-proj-abc123def456ghi789";
const SUPABASE_KEY = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9...";
const DB_PASSWORD = "<YOUR_SECURE_PASSWORD>";
fetch('https://api.example.com', { headers: { Authorization: 'Bearer <YOUR_TOKEN>' } });
// ✅ ALWAYS write code like this:
const API_KEY = process.env.API_KEY;
const SUPABASE_KEY = process.env.SUPABASE_ANON_KEY;
// For Cloudflare Workers:
export default { async fetch(req, env) { const key = env.API_KEY; } };
Layer 2: Pre-Commit Guard — Block Secrets at Commit
Goal: Automatically scan staged files BEFORE they enter git history.
Option A: Gitleaks (Recommended — Industry Standard)
Step 1: Install Gitleaks
# macOS
brew install gitleaks
# Linux
# Download from https://github.com/gitleaks/gitleaks/releases
# Verify installation
gitleaks version
Step 2: Create .gitleaks.toml (Project Root)
# .gitleaks.toml — Cody Master Secret Shield Configuration
title = "CM Secret Shield — Gitleaks Config"
# Extend default rules (catches 100+ known patterns)
[extend]
useDefault = true
# Custom rules for Cody Master projects
[[rules]]
id = "supabase-service-key"
description = "Supabase Service Role Key"
regex = '''eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9\.[a-zA-Z0-9_-]+\.[a-zA-Z0-9_-]+'''
tags = ["supabase", "jwt"]
[[rules]]
id = "cloudflare-api-token"
description = "Cloudflare API Token"
regex = '''[A-Za-z0-9_-]{40}'''
entropy = 4.5
secretGroup = 0
tags = ["cloudflare"]
[[rules]]
id = "generic-high-entropy"
description = "High entropy string that may be a secret"
regex = '''(?i)(api[_-]?key|secret[_-]?key|access[_-]?token|private[_-]?key|auth[_-]?token)\s*[=:]\s*['"][a-zA-Z0-9/+=]{20,}['"]'''
tags = ["generic"]
# Allow patterns (reduce false positives)
[allowlist]
paths = [
'''\.gitleaks\.toml$''',
'''\.dev\.vars\.example$''',
'''node_modules/''',
'''dist/''',
'''\.git/'''
]
Step 3: Setup Git Pre-Commit Hook
# Create the hook
cat > .git/hooks/pre-commit << 'EOF'
#!/bin/bash
# CM Secret Shield — Pre-Commit Guard
# Scans staged files for secrets before allowing commit
echo "🛡️ Secret Shield: scanning staged files..."
# Check if gitleaks is installed
if command -v gitleaks >/dev/null 2>&1; then
gitleaks git --pre-commit --staged --verbose
if [ $? -ne 0 ]; then
echo ""
echo "❌ SECRET DETECTED! Commit blocked."
echo ""
echo "To fix:"
echo " 1. Remove the secret from your code"
echo " 2. Use environment variables instead"
echo " 3. If false positive: add to .gitleaks.toml allowlist"
echo ""
echo "To bypass (DANGEROUS): git commit --no-verify"
exit 1
fi
echo "✅ Secret Shield: no secrets detected"
else
# Fallback: basic pattern check without gitleaks
echo "⚠️ Gitleaks not installed. Running basic checks..."
STAGED=$(git diff --cached --name-only --diff-filter=ACM)
PATTERNS="SERVICE_KEY|ANON_KEY|PRIVATE_KEY|DB_PASSWORD|SECRET_KEY|API_SECRET|sk-[a-zA-Z0-9]{20,}|-----BEGIN.*KEY-----"
for file in $STAGED; do
if echo "$file" | grep -qE '\.(js|ts|jsx|tsx|json|toml|yaml|yml|env|cfg|conf|ini)$'; then
if git diff --cached "$file" | grep -qE "$PATTERNS"; then
echo "❌ Potential secret found in: $file"
echo " Run: git diff --cached $file | grep -E '$PATTERNS'"
exit 1
fi
fi
done
echo "✅ Basic check passed (install gitleaks for deeper scanning)"
fi
EOF
chmod +x .git/hooks/pre-commit
echo "✅ Pre-commit hook installed"
Option B: Native Git Hook Only (No Dependencies)
For projects that can't install Gitleaks, the basic pattern check in the hook above works as a fallback.
Layer 3: Repo Scan — Full Repository Audit
Goal: Scan the entire repository for secrets that may have been committed before the pre-commit hook was set up.
Quick Scan Script
# scripts/security-scan.js — Repo-wide secret detection
node -e "
const fs = require('fs');
const path = require('path');
const DANGEROUS_PATTERNS = [
{ name: 'Service Key Variable', regex: /(?:SERVICE_KEY|SERVICE_ROLE)\s*[=:]\s*['\"][a-zA-Z0-9._\/-]{20,}/g },
{ name: 'Anon Key Variable', regex: /ANON_KEY\s*[=:]\s*['\"][a-zA-Z0-9._\/-]{20,}/g },
{ name: 'Private Key Block', regex: /-----BEGIN\s+(RSA|EC|DSA|OPENSSH)?\s*PRIVATE KEY-----/g },
{ name: 'JWT Token', regex: /eyJ[a-zA-Z0-9_-]{10,}\.[a-zA-Z0-9_-]{10,}\.[a-zA-Z0-9_-]{10,}/g },
{ name: 'Generic API Key', regex: /(?:api[_-]?key|api[_-]?secret|access[_-]?token)\s*[=:]\s*['\"][a-zA-Z0-9\/+=]{20,}['\"/]/gi },
{ name: 'AWS Key', regex: /AKIA[0-9A-Z]{16}/g },
{ name: 'Slack Token', regex: /xox[baprs]-[0-9a-zA-Z-]{10,}/g },
{ name: 'GitHub Token', regex: /gh[ps]_[a-zA-Z0-9]{36,}/g },
{ name: 'Stripe Key', regex: /[sr]k_(test|live)_[a-zA-Z0-9]{20,}/g },
{ name: 'DB Password', regex: /(?:DB_PASSWORD|DATABASE_URL)\s*[=:]\s*['\"][^'\"]{8,}/gi },
];
const SKIP_DIRS = ['node_modules', '.git', 'dist', '.wrangler', '.next', 'coverage'];
const SCAN_EXTS = ['.js', '.ts', '.jsx', '.tsx', '.json', '.toml', '.yaml', '.yml',
'.env', '.cfg', '.conf', '.ini', '.md', '.html', '.jsonc'];
let findings = [];
function scanDir(dir) {
try {
const entries = fs.readdirSync(dir, { withFileTypes: true });
for (const entry of entries) {
if (SKIP_DIRS.includes(entry.name)) continue;
const fullPath = path.join(dir, entry.name);
if (entry.isDirectory()) {
scanDir(fullPath);
} else if (entry.isFile() && SCAN_EXTS.some(ext => entry.name.endsWith(ext))) {
const content = fs.readFileSync(fullPath, 'utf-8');
for (const pattern of DANGEROUS_PATTERNS) {
const matches = content.match(pattern.regex);
if (matches) {
findings.push({ file: fullPath, pattern: pattern.name, count: matches.length });
}
}
}
}
} catch (e) { /* skip unreadable dirs */ }
}
scanDir('.');
if (findings.length > 0) {
console.error('❌ SECRET SCAN FOUND ' + findings.length + ' POTENTIAL ISSUES:');
findings.forEach(f => {
console.error(' ⚠ ' + f.file + ' — ' + f.pattern + ' (' + f.count + ' match(es))');
});
console.error('');
console.error('Actions:');
console.error(' 1. Review each finding — some may be false positives');
console.error(' 2. Move real secrets to .dev.vars (local) or platform secrets (production)');
console.error(' 3. If secret was committed, rotate it IMMEDIATELY');
process.exit(1);
} else {
console.log('✅ Repo scan: no secrets detected in ' + SCAN_EXTS.length + ' file types');
}
"
Add to package.json
{
"scripts": {
"security:scan": "node scripts/security-scan.js",
"security:precommit": "gitleaks git --pre-commit --staged || echo 'Install gitleaks for deep scan'"
}
}
Layer 4: Deploy Gate — Pre-Deploy Secret Audit
Goal: Final check before code leaves the machine. Integrated with
cm-safe-deployGate 0.
Enhanced Gate 0 Check
# Run BEFORE deploy — catches what pre-commit might have missed
node -e "
const fs = require('fs');
// 1. Check tracked files for secrets
const dangerous = ['SERVICE_KEY', 'ANON_KEY', 'DB_PASSWORD', 'SECRET_KEY',
'PRIVATE_KEY', 'API_SECRET', 'AUTH_TOKEN'];
const filesToCheck = [
'wrangler.jsonc', 'wrangler.toml', 'wrangler.json',
'package.json', 'tsconfig.json',
...fs.readdirSync('src').filter(f => f.endsWith('.ts') || f.endsWith('.js')).map(f => 'src/' + f)
].filter(f => fs.existsSync(f));
let failed = false;
for (const file of filesToCheck) {
const content = fs.readFileSync(file, 'utf-8');
for (const key of dangerous) {
// Check for actual values (not just variable names)
const valuePattern = new RegExp(key + '\\\\s*[=:]\\\\s*[\"\\'][a-zA-Z0-9/+=]{20,}', 'g');
if (valuePattern.test(content)) {
console.error('❌ DANGER: ' + file + ' contains a ' + key + ' VALUE');
failed = true;
}
}
}
// 2. Check .gitignore has required patterns
if (fs.existsSync('.gitignore')) {
const gitignore = fs.readFileSync('.gitignore', 'utf-8');
const required = ['.env', '.dev.vars'];
const missing = required.filter(r => !gitignore.includes(r));
if (missing.length > 0) {
console.error('❌ .gitignore missing: ' + missing.join(', '));
failed = true;
}
}
// 3. Check .env files aren't tracked
const { execSync } = require('child_process');
try {
const tracked = execSync('git ls-files', { encoding: 'utf-8' });
const badFiles = ['.env', '.dev.vars', '.env.local', '.env.production'];
const trackedBad = badFiles.filter(f => tracked.split('\\n').includes(f));
if (trackedBad.length > 0) {
console.error('❌ CRITICAL: Secret files tracked by git: ' + trackedBad.join(', '));
console.error(' Fix: git rm --cached ' + trackedBad.join(' '));
failed = true;
}
} catch (e) { /* not a git repo */ }
if (failed) {
console.error('\\n🛡️ Secret Shield: Deploy blocked. Fix issues above.');
process.exit(1);
}
console.log('✅ Secret Shield: deploy gate passed');
"
Layer 5: Runtime Guard — Token Lifecycle Management
Goal: Manage secrets throughout their lifecycle — creation, usage, rotation, revocation.
Token Rotation Schedule
| Platform | Token Type | Max Lifetime | Rotation Trigger |
|---|---|---|---|
| Supabase | anon_key | 90 days | Dashboard → Settings → API |
| Supabase | service_role_key | 30 days | Dashboard → Settings → API |
| Cloudflare | API Token | 90 days | Dashboard → My Profile → API Tokens |
| GitHub | Personal Access Token | 90 days | Settings → Developer Settings → PAT |
| GitHub | Fine-grained Token | 30-90 days | Use expiring tokens when possible |
| OpenAI/Gemini | API Key | 90 days | Rotate in platform dashboard |
Secret Lifecycle File
Track secrets in .secret-lifecycle.json (add to .gitignore!):
{
"_WARNING": "This file tracks secret metadata ONLY. NEVER put actual values here.",
"secrets": [
{
"name": "SUPABASE_ANON_KEY",
"platform": "supabase",
"store": "cloudflare-secrets",
"createdAt": "2026-03-01",
"rotateBy": "2026-06-01",
"lastRotated": "2026-03-01",
"status": "active"
},
{
"name": "SUPABASE_SERVICE_KEY",
"platform": "supabase",
"store": "cloudflare-secrets",
"createdAt": "2026-03-01",
"rotateBy": "2026-04-01",
"lastRotated": "2026-03-01",
"status": "active"
}
]
}
Emergency Rotation Playbook
When a secret is leaked, follow this sequence immediately:
┌─────────────────────────────────────────────────────┐
│ 1. REVOKE — Disable the old key in platform dashboard│
│ 2. ROTATE — Generate a new key │
│ 3. UPDATE — Push new key to secret store │
│ 4. DEPLOY — Redeploy affected services │
│ 5. SCAN — Check git history for the old key │
│ 6. SCRUB — Remove from git history if needed │
│ 7. AUDIT — Review access logs for unauthorized use │
└─────────────────────────────────────────────────────┘
Per-Platform Rotation Commands
Supabase:
# 1. Go to Supabase Dashboard → Project Settings → API
# 2. Click "Regenerate" on the compromised key
# 3. Update Cloudflare Secrets:
wrangler secret put SUPABASE_ANON_KEY # Paste new value
wrangler secret put SUPABASE_SERVICE_KEY # Paste new value
# 4. Update local .dev.vars with new values
# 5. Redeploy
npm run deploy:staging
Cloudflare API Token:
# 1. Dashboard → My Profile → API Tokens → Roll
# 2. Update any CI/CD systems using this token
# 3. Verify with: wrangler whoami
GitHub Token:
# 1. Settings → Developer Settings → PAT → Regenerate
# 2. Update gh auth: gh auth login
# 3. Verify: gh auth status
Security Audit Checklist
Run this checklist for any project to assess its security posture:
## 🛡️ Secret Shield Audit
### Layer 1: Write Guard
- [ ] No hardcoded secrets in source files
- [ ] Environment variables used for all secrets
- [ ] .dev.vars exists with local secrets (not committed)
- [ ] .dev.vars.example exists with placeholder names (committed)
### Layer 2: Pre-Commit Guard
- [ ] .git/hooks/pre-commit exists and is executable
- [ ] Gitleaks installed OR native fallback hook active
- [ ] .gitleaks.toml configured for project
### Layer 3: Repo Scan
- [ ] `npm run security:scan` passes clean
- [ ] No JWT tokens in tracked files
- [ ] No API keys in configuration files
- [ ] No private keys in repository
### Layer 4: Deploy Gate
- [ ] Gate 0 checks ALL source files (not just wrangler.jsonc)
- [ ] .gitignore includes: .env, .dev.vars, .env.local, .env.production
- [ ] No .env files tracked by git
- [ ] Cloudflare Secrets used for production values
### Layer 5: Runtime Guard
- [ ] .secret-lifecycle.json tracks all secrets (metadata only)
- [ ] No secrets past rotation deadline
- [ ] Emergency rotation playbook known by team
- [ ] Post-incident: keys rotated, history scrubbed
Hardened .gitignore Template
Every project using Secret Shield should have AT MINIMUM these patterns:
# === Secret Shield: Mandatory Ignores ===
# Environment & secret files
.env
.env.*
!.env.example
!.env.test
.dev.vars
!.dev.vars.example
# Secret lifecycle tracking (contains metadata, not values)
.secret-lifecycle.json
# Platform-specific
.wrangler/
*.pem
*.key
*.p12
*.pfx
# OS artifacts
.DS_Store
Thumbs.db
# Dependencies
node_modules/
# Build output
dist/
build/
.next/
.nuxt/
# IDE
.vscode/settings.json
.idea/
Red Flags — STOP
| Thought | Reality |
|---|---|
| "It's just a dev key" | Dev keys have the same permissions as prod keys |
| ".gitignore will protect me" | It can't remove what's already in git history |
| "I'll rotate it later" | Later = never. Rotate NOW. |
| "It was only exposed briefly" | Bots scan GitHub in real-time for leaked keys |
| "This is a private repo" | Private doesn't mean secured. Colleagues, CI, forks all have access |
| "The pre-commit hook is annoying" | 3-second scan vs. hours of incident response |
| "I'll add --no-verify just this once" | That "once" is when the leak happens |
Integration
| Skill | Relationship |
|---|---|
cm-project-bootstrap | Phase 0.5 calls Secret Shield for initial security setup |
cm-safe-deploy | Gate 0 uses Layer 4 enhanced secret audit |
cm-test-gate | Layer 5 security test uses Layer 3 patterns |
cm-identity-guard | Layer 5 token rotation extends identity lifecycle |
cm-quality-gate | Secret shield is a prerequisite gate |
Lifecycle Position
cm-project-bootstrap → cm-secret-shield (setup) → development cycle
↓
cm-secret-shield (pre-commit) ← git commit
cm-secret-shield (gate 0) ← cm-safe-deploy
cm-secret-shield (scan) ← cm-test-gate
cm-secret-shield (rotation) ← cm-identity-guard
The Bottom Line
5 layers. Every stage. No exceptions.
Write Guard → Pre-Commit → Repo Scan → Deploy Gate → Runtime Guard.
A 3-second scan prevents a 3-day incident. This is non-negotiable.