* New hook: secrets-scanner Add a secrets-scanner hook that scans files modified during a Copilot coding agent session for leaked secrets, credentials, and sensitive data. The hook runs on sessionEnd and inspects files in one of three scopes: - diff: only files changed in the current session (default) - staged: only files currently staged in the git index - all: every tracked file in the repository Detected pattern categories: - AWS access keys and secret keys - GCP service account credentials - Azure client secrets and storage connection strings - GitHub personal access tokens - Slack tokens (bot, user, webhook) - Private key headers (RSA, EC, DSA, OpenSSH, PEM) - Generic high-entropy bearer tokens - Internal IP:port strings Configurable via environment variables (SCAN_MODE, SCAN_SCOPE, SECRETS_ALLOWLIST) so teams can tune for their workflow without editing the script. Patterns are POSIX ERE (grep -E) compatible, with no PCRE metacharacters, for portability across macOS and Linux. Files: hooks.json, scan-secrets.sh, README.md * refactor: move PATTERNS array to top of scan-secrets.sh for discoverability Move the PATTERNS declaration to the top of the file so it is clearly visible and easy to customize, as suggested in code review. Added a descriptive header comment. No functional changes. --------- Co-authored-by: Shehab Sherif <shehabsherif0@users.noreply.github.com>
7.7 KiB
name, description, tags
| name | description | tags | ||||
|---|---|---|---|---|---|---|
| Secrets Scanner | Scans files modified during a Copilot coding agent session for leaked secrets, credentials, and sensitive data |
|
Secrets Scanner Hook
Scans files modified during a GitHub Copilot coding agent session for accidentally leaked secrets, credentials, API keys, and other sensitive data before they are committed.
Overview
AI coding agents generate and modify code rapidly, which increases the risk of hardcoded secrets slipping into the codebase. This hook acts as a safety net by scanning all modified files at session end for 20+ categories of secret patterns, including:
- Cloud credentials: AWS access keys, GCP service account keys, Azure client secrets
- Platform tokens: GitHub PATs, npm tokens, Slack tokens, Stripe keys
- Private keys: RSA, EC, OpenSSH, PGP, DSA private key blocks
- Connection strings: Database URIs (PostgreSQL, MongoDB, MySQL, Redis, MSSQL)
- Generic secrets: API keys, passwords, bearer tokens, JWTs
- Internal infrastructure: Private IP addresses with ports
Features
- Two scan modes:
warn(log only) orblock(exit non-zero to prevent commit) - Two scan scopes:
diff(modified files vs HEAD) orstaged(git-staged files only) - Smart filtering: Skips binary files, lock files, and placeholder/example values
- Allowlist support: Exclude known false positives via
SECRETS_ALLOWLIST - Structured logging: JSON Lines output for integration with monitoring tools
- Redacted output: Findings are truncated in logs to avoid re-exposing secrets
- Zero dependencies: Uses only standard Unix tools (
grep,file,git)
Installation
-
Copy the hook folder to your repository:
cp -r hooks/secrets-scanner .github/hooks/ -
Ensure the script is executable:
chmod +x .github/hooks/secrets-scanner/scan-secrets.sh -
Create the logs directory and add it to
.gitignore:mkdir -p logs/copilot/secrets echo "logs/" >> .gitignore -
Commit the hook configuration to your repository's default branch.
Configuration
The hook is configured in hooks.json to run on the sessionEnd event:
{
"version": 1,
"hooks": {
"sessionEnd": [
{
"type": "command",
"bash": ".github/hooks/secrets-scanner/scan-secrets.sh",
"cwd": ".",
"env": {
"SCAN_MODE": "warn",
"SCAN_SCOPE": "diff"
},
"timeoutSec": 30
}
]
}
}
Environment Variables
| Variable | Values | Default | Description |
|---|---|---|---|
SCAN_MODE |
warn, block |
warn |
warn logs findings only; block exits non-zero to prevent auto-commit |
SCAN_SCOPE |
diff, staged |
diff |
diff scans uncommitted changes vs HEAD; staged scans only staged files |
SKIP_SECRETS_SCAN |
true |
unset | Disable the scanner entirely |
SECRETS_LOG_DIR |
path | logs/copilot/secrets |
Directory where scan logs are written |
SECRETS_ALLOWLIST |
comma-separated | unset | Patterns to ignore (e.g., test_key_123,example.com) |
How It Works
- When a Copilot coding agent session ends, the hook executes
- Collects all modified files using
git diff(respects the configured scope) - Filters out binary files and lock files
- Scans each text file line-by-line against 20+ regex patterns for known secret formats
- Skips matches that look like placeholders (e.g., values containing
example,changeme,your_) - Checks matches against the allowlist if configured
- Reports findings with file path, line number, pattern name, and severity
- Writes a structured JSON log entry for audit purposes
- In
blockmode, exits non-zero to signal the agent to stop before committing
Detected Secret Patterns
| Pattern | Severity | Example Match |
|---|---|---|
AWS_ACCESS_KEY |
critical | AKIAIOSFODNN7EXAMPLE |
AWS_SECRET_KEY |
critical | aws_secret_access_key = wJalr... |
GCP_SERVICE_ACCOUNT |
critical | "type": "service_account" |
GCP_API_KEY |
high | AIzaSyC... |
AZURE_CLIENT_SECRET |
critical | azure_client_secret = ... |
GITHUB_PAT |
critical | ghp_xxxxxxxxxxxx... |
GITHUB_FINE_GRAINED_PAT |
critical | github_pat_... |
PRIVATE_KEY |
critical | -----BEGIN RSA PRIVATE KEY----- |
GENERIC_SECRET |
high | api_key = "sk-..." |
CONNECTION_STRING |
high | postgresql://user:pass@host/db |
SLACK_TOKEN |
high | xoxb-... |
STRIPE_SECRET_KEY |
critical | sk_live_... |
NPM_TOKEN |
high | npm_... |
JWT_TOKEN |
medium | eyJhbGci... |
INTERNAL_IP_PORT |
medium | 192.168.1.1:8080 |
See the full list in scan-secrets.sh.
Example Output
Clean scan
🔍 Scanning 5 modified file(s) for secrets...
✅ No secrets detected in 5 scanned file(s)
Findings detected (warn mode)
🔍 Scanning 3 modified file(s) for secrets...
⚠️ Found 2 potential secret(s) in modified files:
FILE LINE PATTERN SEVERITY
---- ---- ------- --------
src/config.ts 12 GITHUB_PAT critical
.env.local 3 CONNECTION_STRING high
💡 Review the findings above. Set SCAN_MODE=block to prevent commits with secrets.
Findings detected (block mode)
🔍 Scanning 3 modified file(s) for secrets...
⚠️ Found 1 potential secret(s) in modified files:
FILE LINE PATTERN SEVERITY
---- ---- ------- --------
lib/auth.py 45 AWS_ACCESS_KEY critical
🚫 Session blocked: resolve the findings above before committing.
Set SCAN_MODE=warn to log without blocking, or add patterns to SECRETS_ALLOWLIST.
Log Format
Scan events are written to logs/copilot/secrets/scan.log in JSON Lines format:
{"timestamp":"2026-03-13T10:30:00Z","event":"secrets_found","mode":"warn","scope":"diff","files_scanned":3,"finding_count":2,"findings":[{"file":"src/config.ts","line":12,"pattern":"GITHUB_PAT","severity":"critical","match":"ghp_...xyz1"}]}
{"timestamp":"2026-03-13T10:30:00Z","event":"scan_complete","mode":"warn","scope":"diff","status":"clean","files_scanned":5}
Pairing with Other Hooks
This hook pairs well with the Session Auto-Commit hook. When both are installed, order them so that secrets-scanner runs first:
- Secrets scanner runs at
sessionEnd, catches leaked secrets - Auto-commit runs at
sessionEnd, only commits if all previous hooks pass
Set SCAN_MODE=block to prevent auto-commit when secrets are detected.
Customization
- Add custom patterns: Edit the
PATTERNSarray inscan-secrets.shto add project-specific secret formats - Adjust sensitivity: Change severity levels or remove patterns that generate false positives
- Allowlist known values: Use
SECRETS_ALLOWLISTfor test fixtures or known safe patterns - Change log location: Set
SECRETS_LOG_DIRto route logs to your preferred directory
Disabling
To temporarily disable the scanner:
- Set
SKIP_SECRETS_SCAN=truein the hook environment - Or remove the
sessionEndentry fromhooks.json
Limitations
- Pattern-based detection; does not perform entropy analysis or contextual validation
- May produce false positives for test fixtures or example code (use the allowlist to suppress these)
- Scans only text files; binary secrets (keystores, certificates in DER format) are not detected
- Requires
gitto be available in the execution environment