Disclosure Scan API
Disclosure Scan scores how legibly AI agents will parse a draft filing before it is filed, and returns the exact spans that drive misread risk with a plain fix for each.
What it returns
The response returns a verdict (AI_LEGIBLE, REVIEW_LANGUAGE, or HIGH_MISREAD_RISK), a legibility_score (0 to 100), a factors array over hedging density, quantification, sentiment stability, negation clarity, structural clarity, and extractability, a reasons list, flagged_spans (each with span, risk, and suggestion), a stats block, the trust envelope, and an issuer-facing compliance status.
Each flagged span names the specific parsing hazard, such as a double negative or a stacked hedge, and a concrete remediation.
When to use it
Use Disclosure Scan pre-file, from the CFO or IR office, to see where quant engines and financial agents are likely to misparse hedged, unquantified, or double-negative language. Send the draft text and optionally a symbol.
For the full board-ready report including the institutional simulation and per-tier readiness, use the AI Readiness Report endpoint instead.
Notes and honesty
The scan is deterministic and offline-safe: it counts hedges, numerics, polarity flips, negations, and sentence structure over the supplied text. It records the verdict to the Trust Ledger and feeds the question flywheel. This is an editorial, machine-legibility analysis of the text, never investment advice, a prediction, or a recommendation.
- verdict: AI_LEGIBLE, REVIEW_LANGUAGE, or HIGH_MISREAD_RISK
- legibility_score: 0 to 100
- factors: hedging, quantification, sentiment, negation, structure, extractability
- flagged_spans: the risky sentences plus a fix each
- stats: word, sentence, hedge, and numeric counts
- ledger_id: the recorded verdict for audit
curl -s https://api.lyramindos.com/v1/disclosure/scan -H 'Content-Type: application/json' -H 'x-api-key: YOUR_KEY' -d '{"text":"Results were not inconsistent with prior guidance, and margins may improve.","symbol":"ACME"}'Questions
Who is this for?
The buyer is the CFO or IR office, not the retail investor. It is a pre-file check on how machine-readable a draft disclosure is before AI agents and aggregators parse it.
Does it claim to game a specific model?
No. It is an editorial, machine-legibility analysis of the text itself, never a claim about how any named model will respond, and never advice or a prediction.
How is it different from AI Readiness Report?
Scan is the legibility primitive: verdict, factors, and flagged spans. The report adds the institutional simulation, per-tier readiness dimensions, predicted questions, and recommendations.