Goal
Each example under `examples/` currently ships one passing `input.json`. Add failing-case variants so newcomers can see what a non-compliant verdict looks like without editing the passing input.
Why this matters
Failing-case fixtures teach Rego deny paths and make the policy's actual guards obvious. They're also the easiest "first PR" for a new contributor.
Per-example additions
- `input-low-completeness.json` — documentation completeness below 0.7
- `input-high-toxicity.json` — toxicity above 0.7
- `input-accountability-fail.json`, `input-transparency-fail.json`, `input-fairness-fail.json` — one per sub-function failure
- `input-above-threshold.json` — toxicity_score above the configured threshold
Acceptance criteria
Get help
Comment here. Splitting this across 3 PRs (one per example) is welcome.
Goal
Each example under `examples/` currently ships one passing `input.json`. Add failing-case variants so newcomers can see what a non-compliant verdict looks like without editing the passing input.
Why this matters
Failing-case fixtures teach Rego deny paths and make the policy's actual guards obvious. They're also the easiest "first PR" for a new contributor.
Per-example additions
examples/eu-ai-act-transparency/
examples/nist-ai-rmf-govern/
examples/customer-support-llm/
Acceptance criteria
Get help
Comment here. Splitting this across 3 PRs (one per example) is welcome.