haluguard/haluguard
Verify LLM output for hallucinations using haluguard.com
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| token | haluguard API token (get one at haluguard.com) | yes | — |
| text | Text content to verify (alternative to file). | no | — |
| file | Path to a file whose contents will be verified. | no | — |
| threshold | Hallucination score 0..1; if score >= threshold, the action fails. Default 0.7. Pass "0" to fail on any hallucination. | no | 0.7 |
| fail-on-hallucinations | If "true", any "hallucinations_found" verdict fails the build (overrides threshold). Default "false". | no | false |
| api-url | Override the API base URL (for self-hosted / dev). | no | https://api.haluguard.com |
Outputs
| name | description |
|---|---|
| verdict | "clean" or "hallucinations_found" |
| severity | severity tier: clear / suspicious / unclear / clean |
| score | overall score 0..1 |
| n_claims | total claims found in the text |
| n_hallucinations | claims flagged as hallucinations |
| report | full JSON report from haluguard |