AI and memory

aiReview.correct

Append an evidence-backed human correction bound to the exact conversation snapshot. Sentiment stays separate from accuracy. Requires correct and read grants for this exact owner. Human dashboard or terminal only.

POST/api/v1/accounting/aiReview.correct

Permissions

accounting:readaccounting:write

Who can call it

admin, finance, contributor

Retries

No idempotency key

MCP

Not offered over MCP: Private redacted chats and human review stay in Oatmilk and its terminal; connected AI apps receive no reviewer access or chat sharing.

Fields

  • sessionIdstringRequired

    Matches ^[A-Za-z0-9_-]{1,200}$

  • snapshotHashstringRequired

    SHA-256 hash as 64 lowercase hex characters

  • expectedAnswerstringRequired

    1–10000 characters

  • assessmentsarray of objectsRequired

    at most 50 items

  • calculationsarray of objects

    at most 20 items · Default []

  • sentimentenum
    positivenegativeneutralunknown

    Default "unknown"

Example

curl https://app.getoatmilk.com/api/v1/accounting/aiReview.correct \
  -H "Authorization: Bearer $OATMILK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "sessionId": "session_id",
  "snapshotHash": "9f2b5c1d7e3a4b6c8d0e2f4a6b8c0d2e4f6a8b0c2d4e6f8a0b2c4d6e8f0a2b4c",
  "expectedAnswer": "example",
  "assessments": [
    {
      "dimension": "groundedAccuracy",
      "verdict": "pass",
      "messageId": "message_id",
      "evidence": "example",
      "rationale": "example"
    }
  ]
}'
Response
{
  "data": { … }
}

Try it

Try it

Checks your input with this action’s real schema and answers like the API, with synthetic data. No key needed, and nothing changes.

POST/api/v1/accounting/aiReview.correct
curl https://app.getoatmilk.com/api/v1/accounting/aiReview.correct \
  -H "Authorization: Bearer $OATMILK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "sessionId": "session_id",
  "snapshotHash": "9f2b5c1d7e3a4b6c8d0e2f4a6b8c0d2e4f6a8b0c2d4e6f8a0b2c4d6e8f0a2b4c",
  "expectedAnswer": "example",
  "assessments": [
    {
      "dimension": "groundedAccuracy",
      "verdict": "pass",
      "messageId": "message_id",
      "evidence": "example",
      "rationale": "example"
    }
  ]
}'

More in AI and memory.