evidence_for

See the evidence trail behind a belief — every supporting or refuting observation, with strengths and timestamps. Use this to answer "why is this belief at confidence X?" instead of taking the number on faith.

Signature

evidence_for({ belief_id: number }) → {
  belief: { id, text, confidence, sprt_log_ratio, evidence_count, sprt_status, last_evidence_at },
  events: Array<{ id, kind, strength, log_likelihood, note, created_at }>
}
ParamTypeRequiredDescription
belief_idintyesId of the belief to inspect.

Example

curl "https://mneva.dev/v1/evidence/1" \
  -H "x-mneva-key: $MNEVA_KEY"

Response:

{
  "belief": {
    "id": 1,
    "text": "the worker pool should be sized at 4 threads",
    "confidence": 0.4,
    "sprt_log_ratio": -5.2038,
    "evidence_count": 3,
    "sprt_status": "demoted",
    "last_evidence_at": "2026-05-23T00:10:06.078Z"
  },
  "events": [
    {
      "id": 1,
      "kind": "refuting",
      "strength": 0.85,
      "log_likelihood": -1.7346,
      "note": "observation 1: production needed 8 threads under load",
      "created_at": "2026-05-23T00:10:06.036Z"
    },
    {
      "id": 2,
      "kind": "refuting",
      "strength": 0.85,
      "log_likelihood": -1.7346,
      "note": "observation 2: production needed 8 threads under load",
      "created_at": "2026-05-23T00:10:06.056Z"
    },
    {
      "id": 3,
      "kind": "refuting",
      "strength": 0.85,
      "log_likelihood": -1.7346,
      "note": "observation 3: production needed 8 threads under load",
      "created_at": "2026-05-23T00:10:06.078Z"
    }
  ]
}

The belief's confidence is 0.4 because three refuting observations at strength 0.85 pushed the SPRT log ratio past -2.25 on the second one, demoting confidence by 0.20 from 0.6.

The third event is on the record (audit completeness) but did not move confidence further — once demoted, the belief stays there until revise is called.

See also

Was this page helpful?