| mode | Operating mode: "check" (default) runs gates on PR, "baseline" generates baseline.json | no | check |
| fail_fast | Stop running gates after first blocking gate failure | no | true |
| gates | Comma-separated list of gates to run (typescript,eslint,semgrep,gitleaks,npm-audit,design-system,visual,llm-review) | no | typescript,eslint,semgrep,gitleaks |
| config_path | Path to Hawky configuration file | no | .hawky.yml |
| github_token | GitHub token for PR comments (defaults to GITHUB_TOKEN) | no | ${{ github.token }} |
| commit_baseline | Commit and push baseline files after generation (only for baseline mode) | no | false |
| visual_enabled | Enable visual regression testing | no | false |
| visual_threshold | Threshold for visual differences (0.0-1.0, where 1.0 = 100% different) | no | 0.1 |
| visual_routes | Comma-separated list of routes to test visually | no | "" |
| stacks | Comma-separated list of stacks to validate (typescript,go,rust,python,terraform,docker,kubernetes) or "auto" for auto-detection | no | auto |
| llm_enabled | Enable LLM-powered code review | no | false |
| llm_api_key | API key for LLM service (Azure AI Foundry or compatible) | no | "" |
| llm_endpoint | LLM endpoint URL (Azure AI Foundry deployment URL) | no | "" |
| backend_checks_enabled | Enable backend convention checks (NestJS repository pattern, DTO validation, N+1 queries, unbounded queries) | no | false |
| sprint_enabled | Enable sprint tracking integration | no | false |
| sprint_file | Path to sprint file for integration | no | .claude/work/SPRINT.md |
| coordination_enabled | Enable cross-agent coordination checks | no | true |
| stale_branch_threshold | Days before marking a branch as stale for coordination checks | no | 10 |