Agent Readiness API
Readiness is the go/no-go gate an agent checks before it acts. Given a symbol and an optional action it returns AGENT_READY, REVIEW_REQUIRED, or DO_NOT_EXECUTE plus the factors behind the verdict.
What it returns
The response carries a verdict, a readiness_score (0 to 100), and a factors array over grounding, freshness, confidence, signal agreement, contradiction, and hallucination risk, each with a pass/warn/fail status and detail. It also returns reasons, the plain market_read, institution_memory (the name's Trust Ledger track record), a verification summary (consensus, contradictions, reasoners, models_live), and the trust envelope.
Each call is written to the Trust Ledger and the entry id is surfaced as ledger_id for later reconciliation.
When to use it
Call Readiness right before an agent would execute on a symbol. It composes the Market Brain read, the confluence signal, the multi-reasoner verification panel, and Institution Memory into one decision gate. A weak track record can only tighten a verdict, never loosen it.
The optional action field labels what the agent is about to do and is echoed back and recorded.
Notes and honesty
The verdict downgrades honestly: ungrounded data, confidence below 0.4, or two or more contradictions force DO_NOT_EXECUTE; sample data, low confidence, weak agreement, or a single contradiction force REVIEW_REQUIRED. This is a data-readiness gate, not a recommendation to buy, sell, or hold.
- verdict: AGENT_READY, REVIEW_REQUIRED, or DO_NOT_EXECUTE
- readiness_score: 0 to 100 composite
- factors: grounding, freshness, confidence, agreement, contradiction, hallucination risk
- verification: consensus, contradictions, reasoners, models_live
- institution_memory: the name's Trust Ledger track record
- ledger_id: the recorded entry for later outcome reconciliation
curl -s https://api.lyramindos.com/v1/readiness -H 'Content-Type: application/json' -H 'x-api-key: YOUR_KEY' -d '{"symbol":"AMD","action":"rebalance"}'Questions
How does the verdict get decided?
It composes grounding, freshness, confidence, signal agreement, contradictions from the verification panel, and hallucination risk. Ungrounded data or two contradictions force DO_NOT_EXECUTE; sample data or weak agreement force REVIEW_REQUIRED.
Is this a buy signal?
No. It is a data-readiness gate that judges whether the intelligence is safe to act on, never a recommendation to buy, sell, or hold.
What is ledger_id for?
Every verdict is written to the Trust Ledger. The returned ledger_id lets you later reconcile the outcome via POST /v1/ledger/outcome, which sharpens the name's track record.