Detecting exposed AWS credentials in source code repositories, CI/CD pipelines, and configuration files using TruffleHog, git-secrets, and AWS-native detection mechanisms to prevent credential theft and unauthorized account access.
Detecting exposed AWS credentials in source code repositories, CI/CD pipelines, and configuration files using TruffleHog, git-secrets, and AWS-native detection mechanisms to prevent credential theft and unauthorized account access.
When integrating secrets detection into CI/CD pipelines to prevent credential commits reaching production
When performing a security audit of existing repositories for historically committed AWS credentials
When responding to an AWS GuardDuty alert about credential usage from an unexpected IP or region
When onboarding repositories from acquired companies or third-party vendors
When validating that credential rotation processes have removed all references to old access keys
Do not use for real-time credential monitoring (use AWS GuardDuty or Amazon Macie), for managing secrets (use AWS Secrets Manager or HashiCorp Vault), or for detecting non-credential sensitive data like PII (use Amazon Macie or DLP tools).
Detection Gaps & Validation
--only-verified suppresses real leaks: verification calls sts:GetCallerIdentity, so a key that is valid-but-disabled, in an account you can't reach, or rate-limited gets dropped as "unverified." Run once without--only-verified and triage the UNVERIFIED count too — a leaked-then-deactivated key still proves a process failure.
Default scan is shallow:trufflehog git without full depth scans only reachable history of the current ref. Force-pushed/rewritten commits, git stash, dangling blobs, other branches, and forks can retain secrets. Scan all branches and consider --max-depth/repo mirror clones.
Secrets live outside git: CI/CD logs, container image layers, S3 buckets, Terraform state, and .env artifacts — use the filesystem, s3, and CI integrations, not just git.
git-secrets only catches AKIA… long-term keys: temporary ASIA… STS tokens and non-AWS/custom formats slip past the default --register-aws patterns; add explicit regexes.
Validate findings against AWS: for every AKIA/ASIA hit, run aws iam get-access-key-last-used --access-key-id … and aws iam list-access-keys to confirm Active status and whether it was used (CloudTrail by AccessKeyId) before declaring it benign or expired.
Prerequisites
TruffleHog v3 installed (brew install trufflehog or pip install trufflehog)
git-secrets installed for pre-commit hook integration (brew install git-secrets)
Access to source code repositories (GitHub, GitLab, Bitbucket, or local git repos)
AWS CLI configured with permissions to check key status (iam:ListAccessKeys, iam:GetAccessKeyLastUsed)
GitHub or GitLab API token for scanning organization-wide repositories
Workflow
Step 1: Install and Configure TruffleHog
Install TruffleHog v3 and verify it can detect the AWS credential patterns.
# Install TruffleHog v3
pip install trufflehog
# Or install from binary release
curl -sSfL https://raw.githubusercontent.com/trufflesecurity/trufflehog/main/scripts/install.sh | sh -s -- -b /usr/local/bin
# Verify installation
trufflehog --version
# Test with a known test repository
trufflehog git https://github.com/trufflesecurity/test_keys --only-verified
Step 2: Scan Git Repositories for Exposed Credentials
Scan entire git history including all branches and commits for AWS access keys, secret keys, and session tokens.
# Scan a local git repository (full history)
trufflehog git file:///path/to/repo --only-verified --json > trufflehog-results.json
# Scan a GitHub organization's repositories
trufflehog github --org=your-organization --token=$GITHUB_TOKEN --only-verified
# Scan a specific GitHub repository with all branches
trufflehog git https://github.com/org/repo.git --only-verified --branch=main
# Scan a GitLab group
trufflehog gitlab --group=your-group --token=$GITLAB_TOKEN --only-verified
# Scan filesystem paths for credentials in config files
trufflehog filesystem /path/to/project --only-verified
Step 3: Analyze and Validate Detected Credentials
Parse TruffleHog results to identify verified (still-active) credentials versus rotated or test keys.
# Parse TruffleHog JSON output for AWS findingscat trufflehog-results.json | python3 -c "
import json, sys
for line in sys.stdin:
finding = json.loads(line)
if 'AWS' in finding.get('DetectorName', ''):
print(f\"Detector: {finding['DetectorName']}\")
print(f\"Verified: {finding.get('Verified', False)}\")
print(f\"Source: {finding.get('SourceMetadata', {})}\")
print(f\"Commit: {finding.get('SourceMetadata', {}).get('Data', {}).get('Git', {}).get('commit', 'N/A')}\")
print(f\"File: {finding.get('SourceMetadata', {}).get('Data', {}).get('Git', {}).get('file', 'N/A')}\")
print('---')
"# Check if a detected access key is still active
aws iam get-access-key-last-used --access-key-id AKIAIOSFODNN7EXAMPLE
# List all access keys for a user to find active keys
aws iam list-access-keys --user-name target-user \
--query 'AccessKeyMetadata[*].[AccessKeyId,Status,CreateDate]' --output table
Step 4: Set Up Pre-Commit Hooks with git-secrets
Prevent credentials from being committed in the first place using git-secrets as a pre-commit hook.
# Install git-secrets
git secrets --install # In each repository# Register AWS credential patterns
git secrets --register-aws
# Add custom patterns for internal credential formats
git secrets --add 'AKIA[0-9A-Z]{16}'
git secrets --add 'aws_secret_access_key\s*=\s*.{40}'
git secrets --add 'aws_session_token\s*=\s*.+'# Scan entire repository history
git secrets --scan-history
# Add to global git template for all new repos
git secrets --install ~/.git-templates/git-secrets
git config --global init.templateDir ~/.git-templates/git-secrets
Step 5: Integrate TruffleHog into CI/CD Pipeline
Add TruffleHog scanning as a CI/CD gate to block deployments containing exposed credentials.
# GitLab CI (.gitlab-ci.yml)secrets_scan:stage:testimage:trufflesecurity/trufflehog:latestscript:-trufflehoggitfile://$CI_PROJECT_DIR--since-commit$CI_COMMIT_BEFORE_SHA--only-verified--failallow_failure:false
Step 6: Respond to Detected Credential Exposure
Execute incident response procedures when verified credentials are found exposed.
# IMMEDIATE: Deactivate the exposed access key
aws iam update-access-key \
--user-name compromised-user \
--access-key-id AKIAEXPOSEDKEY123456 \
--status Inactive
# Generate new credentials
aws iam create-access-key --user-name compromised-user
# Review CloudTrail for unauthorized usage of the exposed key
aws cloudtrail lookup-events \
--lookup-attributes AttributeKey=AccessKeyId,AttributeValue=AKIAEXPOSEDKEY123456 \
--start-time 2026-01-01T00:00:00Z \
--query 'Events[*].[EventTime,EventName,EventSource,SourceIPAddress]' \
--output table
# Delete the exposed key after rotation is confirmed
aws iam delete-access-key \
--user-name compromised-user \
--access-key-id AKIAEXPOSEDKEY123456
# Remove the credential from git history using BFG Repo Cleaner
java -jar bfg.jar --replace-text credentials.txt repo.git
Key Concepts
Term
Definition
TruffleHog
Open-source secrets detection tool that scans git history, filesystems, and cloud services for exposed credentials using regex patterns and verification APIs
Verified Secret
A credential that TruffleHog has confirmed is still active by making an API call to the target service (e.g., AWS STS GetCallerIdentity)
git-secrets
AWS Labs pre-commit hook tool that prevents committing strings matching AWS credential patterns to git repositories
Access Key Rotation
The practice of regularly replacing AWS access key pairs to limit the window of exposure if a key is compromised
BFG Repo Cleaner
Tool for removing sensitive data from git history without rewriting the entire repository, faster than git filter-branch
GitHub Secret Scanning
GitHub-native feature that scans public repositories for known credential patterns and notifies the credential provider
Tools & Systems
TruffleHog v3: Primary scanning engine supporting git, filesystem, S3, and CI/CD integration with verified credential detection
git-secrets: AWS Labs pre-commit hook for preventing credential commits at the developer workstation level
BFG Repo Cleaner: Fast tool for removing credentials from git history after exposure is detected
AWS GuardDuty: Threat detection service that alerts on anomalous usage of AWS credentials from unexpected locations
GitHub Advanced Security: Platform-native secret scanning for GitHub repositories with push protection
Common Scenarios
Scenario: Developer Commits AWS Credentials to a Public GitHub Repository
Context: GitHub secret scanning notifies that an AWS access key was pushed to a public repository. The key belongs to a developer with production S3 and DynamoDB access.
Approach:
Immediately deactivate the access key using aws iam update-access-key --status Inactive
Run aws cloudtrail lookup-events filtering by the exposed AccessKeyId to check for unauthorized usage
Scan the full repository history with trufflehog git to find any other exposed credentials
Generate a new access key for the developer and deliver it through Secrets Manager
Remove the credential from git history using BFG Repo Cleaner
Install git-secrets pre-commit hook on the developer's workstation
Add TruffleHog to the repository's CI/CD pipeline to prevent recurrence
Pitfalls: Simply deleting the commit or force-pushing does not remove credentials from GitHub's cache or forks. The key must be deactivated at the AWS level immediately. GitHub secret scanning may have already notified AWS, triggering automated key deactivation.
Output Format
AWS Credential Exposure Scan Report
======================================
Scan Target: github.com/acme-corp (42 repositories)
Scan Date: 2026-02-23
Tool: TruffleHog v3.63.0
Mode: Full git history scan with verification
VERIFIED FINDINGS (Active Credentials):
[CRED-001] AWS Access Key - VERIFIED ACTIVE
Key ID: AKIA...WXYZ
Repository: acme-corp/backend-api
File: deploy/config.env
Commit: a1b2c3d (2025-08-15)
Author: developer@acme.com
IAM User: svc-backend-deploy
Permissions: S3, DynamoDB, SQS (production)
Status: CRITICAL - Key active and used from 3 IP addresses
Action Required: Immediate deactivation and rotation
[CRED-002] AWS Secret Key - VERIFIED ACTIVE
Repository: acme-corp/data-pipeline
File: scripts/etl_config.py
Commit: d4e5f6g (2025-11-22)
Author: data-engineer@acme.com
Status: HIGH - Key active, last used 2 days ago
UNVERIFIED FINDINGS (Potential Credentials):
Total pattern matches: 15
Likely test/example keys: 12
Requires manual review: 3
SUMMARY:
Repositories scanned: 42
Commits analyzed: 125,847
Verified active credentials: 2
Unverified credential patterns: 15
Repositories with pre-commit hooks: 8 / 42
