| github-token | GitHub token for API access and GitHub Models | no | ${{ github.token }} |
| mode | scan (free, report only) or fix (BYOK, auto-fix + PR) | no | "" |
| agent | AI agent to use: claude or codex | no | claude |
| timeout | Max run time (e.g. 30m, 2h, 5h50m) | no | 30m |
| max-cost | Max API spend per run (e.g. $5.00) | no | $5.00 |
| max-passes | Max scan/fix passes before stopping | no | 10 |
| min-passes | Minimum passes to confirm repo is clean | no | 2 |
| max-chains | Max self-continuations for long runs (6h each) | no | 3 |
| strictness | Issue detection strictness: low, medium, high | no | high |
| fix-types | Comma-separated issue types to fix | no | security,bugs,types,lint,dead-code,stubs,duplicates,coverage |
| model | Override AI model (e.g. claude-sonnet-4-5-20250929) | no | "" |
| github-models-model | Model for scan via GitHub Models (free: openai/gpt-4o-mini, premium: openai/gpt-4o, openai/o1-mini, etc.) | no | openai/gpt-4o-mini |
| test-command | Custom test command (auto-detected if empty) | no | "" |
| gist-id | GitHub Gist ID for dynamic badge updates | no | "" |
| gist-token | PAT with gist scope for badge updates | no | "" |
| fail-below | Fail the action if score drops below this | no | 0 |
| verbose | Stream agent output to the Actions log in real-time | no | false |
| max-turns | Max agent turns per invocation (scan: 30, fix: 15) | no | "" |
| max-issues-per-pass | Max issues to fix per pass (0 = unlimited) | no | 0 |
| output-file | Write full issues JSON to this path (e.g. .sloppy/issues.json). Empty to disable. | no | "" |
| scan-scope | What to scan: auto (PR files on pull_request, full otherwise), pr (PR files only), full (entire repo) | no | auto |
| scan-provider | Which AI provider to use for scans: github-models (free tier) or agent (same provider/model as fix mode, requires API key) | no | github-models |
| custom-prompt | Custom instructions injected into every scan/fix prompt (inline text) | no | "" |
| custom-prompt-file | Path to a file containing custom prompt instructions (relative to repo root) | no | "" |
| plugins | Enable/disable the plugin system (loads from .sloppy/plugins/) | no | true |
| parallel-agents | Number of parallel agents to dispatch for fixing (1-8). Uses git worktrees for true parallelism. | no | 1 |
| min-severity | Minimum issue severity to report/fix: critical, high, medium, low. Issues below this level are ignored. | no | low |
| profile | Load a config profile from .sloppy/profiles/<name>.yml. Profile values override .sloppy.yml. | no | "" |
| chain_number | Internal: chain continuation number | no | 0 |