analyze_cognitive

Server-side AI cognitive analysis - the BILLED, opt-in FALLBACK to local scoring. Rates the cognitive difficulty of candidate keys and suggests plain-language rewrites server-side. Prefer scoring locally with set_cognitive_score (0 credits); use this only for very large volumes or on explicit request. Bills credits (standard=1, premium=4 per key; 402 if insufficient, 503 if AI unavailable). Returns {requested, analyzed, credits_charged, credits_refunded}.

Paramètres

Nom	Type		Description
project_uuid	string	optionnel	Project UUID. Required when multiple projects are configured (SONENTA_PROJECTS); defaults to the single configured project otherwise.
tier	string	optionnel	Quality tier (credit_cost per key: standard=1, premium=4). (standard, premium)
key_uuids	array	optionnel	Restrict to these keys (from list_cognitive_candidates). Omit to scope the whole project (or a namespace).
namespace_uuid	string	optionnel	Restrict to one namespace by its UUID.
idempotency_key	string	optionnel	Stable token making this billable call safe to retry: reuse the SAME value to avoid a double-charge. A random one is generated if omitted.

Exemple

{
  "tool": "analyze_cognitive",
  "arguments": {}
}