designer-coderajay/Glassbox EU AI Act Compliance Audit
EU AI Act Annex IV compliance gate for CI/CD. Fails the build if model explainability grade falls below your threshold.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Activelast commit Jun 28, 2026
- License
- MIT
Pinned Snippet
uses: designer-coderajay/glassbox-mech@3942b5cb81096f03b55fe25cae63d67357fef3ea # v4.4.0tags can be moved; commit SHAs can't. why a SHA?
Inputs
| name | description | required | default |
|---|---|---|---|
| api_url | Glassbox API endpoint. Use the public hosted API or your self-hosted instance. | no | https://designer-coderajay-glassbox-ai-2-0-mechanistic-interpretability-tool.hf.space |
| model_name | Model to audit (TransformerLens-compatible for white-box, e.g. gpt2, gpt2-medium) | yes | — |
| prompt | Decision prompt for attribution analysis (the input your model makes decisions on) | yes | — |
| correct_token | Expected correct output token (e.g. " Mary", " approved") | yes | — |
| incorrect_token | Distractor / incorrect token (e.g. " John", " denied") | yes | — |
| provider_name | Provider name for Annex IV Section 1 (your company/team name) | no | CI Pipeline |
| provider_address | Provider address for Annex IV Section 1 | no | "" |
| system_purpose | Purpose of the AI system (for Annex IV documentation) | no | Automated decision-making system under EU AI Act review |
| deployment_context | Deployment context. One of: financial_services | healthcare | hr_employment | legal | critical_infrastructure | education | other_high_risk | no | other_high_risk |
| method | Attribution method. "taylor" (fast, default) or "integrated_gradients" (more accurate) | no | taylor |
| fail_below_grade | Fail the CI step if the explainability grade is strictly below this. A = best, D = worst. Set to "D" to never fail on grade. | no | C |
| output_path | File path to write the JSON compliance report | no | glassbox-report.json |
Outputs
| name | description |
|---|---|
| grade | Explainability grade: A (fully explainable) → D (not explainable) |
| f1_score | Faithfulness F1 score (0.0–1.0) |
| compliance_status | Compliance status: conditionally_compliant | incomplete | non_compliant |
| report_id | Glassbox report ID — use to download PDF/JSON from the API |
| sufficiency | Faithfulness sufficiency score |
| comprehensiveness | Faithfulness comprehensiveness score |