Pipeline Auditor
Security and DevOps specialist focused on auditing CI/CD pipelines, deployment configurations, and infrastructure as code.Quick Reference
| Property | Value |
|---|---|
| Domain | DevOps |
| FORGE Stage | 5 (VERIFY) |
| Version | 1.0.0 |
| Output Types | Audit reports (JSON, Markdown) |
Overview
Use this agent when you need to:
- Audit GitHub Actions workflows for security vulnerabilities
- Review Railway deployment configurations
- Identify performance inefficiencies in CI/CD pipelines
- Ensure compliance with security best practices
- Generate remediation checklists with prioritized actions
- Validate supply chain security (pinned actions, dependencies)
Core Capabilities
Security Audit
Identify vulnerabilities in CI/CD configs (secrets, permissions, supply chain)
Performance Analysis
Find inefficiencies in build/deploy pipelines (caching, parallelization)
Best Practice Review
Check compliance with DevOps standards and industry recommendations
Remediation Planning
Provide prioritized, actionable fix recommendations with code examples
When to Use
Auditing security of GitHub Actions workflows
Reviewing Railway deployment configurations
Identifying exposed secrets or hardcoded credentials
Validating supply chain security (action pinning)
Finding performance optimization opportunities
Ensuring SOC 2 or SLSA compliance
Usage Examples
- Security Audit
- Performance Analysis
- Remediation Checklist
Comprehensive security audit of GitHub Actions workflows:Result: Detailed security findings with CWE mappings, evidence, and immediate action items.
Outputs
Audit Report Structure
Severity Definitions
| Severity | Description | Response Time |
|---|---|---|
| Critical | Active exploit possible, data exposure | 24 hours |
| High | Security weakness, requires action | 1 week |
| Medium | Should be fixed, lower risk | 2 weeks |
| Low | Best practice violation | 1 month |
| Informational | Suggestion for improvement | As capacity allows |
Compliance Status
FORGE Gate Compliance
Entry Gates (Pre-conditions)
Target repository identified
Target repository identified
The repository or configuration to be audited must be specified with appropriate access permissions.
Audit scope defined
Audit scope defined
The audit type (security, performance, or full) and specific files/workflows to analyze must be determined.
Access to configurations granted
Access to configurations granted
Read access to
.github/workflows/, railway.json, Dockerfile, and related infrastructure files.Exit Gates (Post-conditions)
Audit report generated
Audit report generated
A complete audit report with findings, severity ratings, and evidence is produced in JSON format.
Vulnerabilities categorized
Vulnerabilities categorized
All security vulnerabilities are categorized by severity (critical, high, medium, low) with CWE mappings.
Remediation steps provided
Remediation steps provided
Each finding includes specific, actionable remediation steps with code examples where applicable.
Compliance status summarized
Compliance status summarized
If compliance frameworks (SOC 2, SLSA) are relevant, gaps and current status are documented.
Integration Points
Control Plane API
Veritas Prompts
- Consumed Prompts
- Produced Prompts
| Prompt ID | Purpose |
|---|---|
vrt-i9j0k1l2 | Security audit checklist (comprehensive security review criteria) |
vrt-m3n4o5p6 | Performance optimization patterns (CI/CD efficiency techniques) |
Audit Checklists
GitHub Actions Security
| Check | Severity | Description |
|---|---|---|
| Secrets in code | Critical | No hardcoded secrets in workflows |
| Permissions | High | Least-privilege permissions used |
| Action pinning | High | Actions pinned to SHA, not tags |
| Third-party actions | Medium | Trusted publishers only |
| Workflow triggers | Medium | No dangerous trigger combinations (e.g., pull_request_target with code execution) |
| Environment protection | Medium | Production requires approval |
| OIDC for cloud | Medium | Use OIDC instead of long-lived tokens |
| Artifact handling | Low | No sensitive data in artifacts |
Railway Security
| Check | Severity | Description |
|---|---|---|
| Variable exposure | High | No secrets in railway.json |
| Network policy | Medium | Internal services not publicly exposed |
| Resource limits | Medium | Memory/CPU limits set appropriately |
| Health checks | Medium | Health endpoints configured |
| Replica count | Low | HA (2+ replicas) for production services |
Container Security
| Check | Severity | Description |
|---|---|---|
| Base image | High | Official, minimal base image used |
| Root user | High | Non-root user configured |
| Secrets in image | Critical | No secrets baked into Dockerfile |
| Layer optimization | Low | Multi-stage builds used |
| Vulnerability scan | Medium | No critical CVEs in dependencies |
Related Agents
| Agent | Relationship | Use Case |
|---|---|---|
| GitHub Actions Engineer | Audits | Reviews generated workflows for security issues |
| Railway Deployer | Audits | Reviews deployment configs for vulnerabilities |
| Incident Commander | Escalates | Receives alerts for critical security findings |
| Runbook Writer | Documents | Creates remediation runbooks from audit findings |
Source Files
View Agent Source
Repository:
Path:
Version: 1.0.0
so1-io/so1-agentsPath:
agents/devops/pipeline-auditor.mdVersion: 1.0.0
Common Patterns
Finding Categories
Security (SEC-xxx):- Exposed secrets (hardcoded, in logs, in history)
- Overly permissive permissions
- Supply chain risks (unpinned actions, untrusted publishers)
- Insecure workflow triggers
- Missing dependency caching
- Sequential jobs that could be parallel
- Inefficient resource allocation
- Redundant build steps
- Missing concurrency controls
- No timeout configurations
- Insufficient error handling
- Missing documentation
Common Audit Failures
| Error | Cause | Resolution |
|---|---|---|
| Access denied | Missing repo permissions | Request read access to repository |
| File not found | Workflow path incorrect | Verify .github/workflows/ directory exists |
| Parse error | Invalid YAML syntax | Fix YAML syntax before running audit |
| Rate limit | Too many API calls | Implement rate limiting or use GitHub App |
Success Metrics
| Metric | Target | Measurement |
|---|---|---|
| Finding Accuracy | >95% | Valid findings / total findings |
| Coverage | 100% | Files analyzed / files in scope |
| Remediation Rate | >80% | Fixed findings / total findings (30 days) |
| False Positive Rate | <5% | False positives / total findings |
Next Steps: