Clinical evidence-based longevity consultation engine using RAG (Retrieval-Augmented Generation) with semantic search over 1100+ medical knowledge chunks
Single tool with minimal definition quality. Tool name 'query_longevity_expert' lacks action verb prefix (should be 'search_' or 'query_'). Description is present but generic (48 chars, below 50-200 char baseline for LLM optimization). Input schema exists with one parameter 'query' (string, described), but output schema is completely undocumented, tool returns a raw string prompt_guide rather than structured data. No error handling, no validation, no recovery guidance. Tool does not follow pattern:tool or pattern:response-shaper. The implementation returns a prompt template string rather than a structured clinical report, forcing LLMs to parse unstructured text.
Search clinical evidence and return a structured report.
No output schema documented. Tool returns raw string (prompt_guide) instead of structured object with fields like overview, mechanism, protocol, goals. LLMs cannot extract or chain results.
Tool name lacks action verb. 'query_longevity_expert' is vague; should be 'search_clinical_evidence' or 'query_health_knowledge_base' to clarify intent.
Description is too short (48 chars, below 50-char minimum for LLM optimization). Does not explain WHEN to use, WHAT it returns, or any prerequisites. Missing dependency hints.
No error handling or recovery guidance. If query is empty, malformed, or returns no results, tool provides no actionable error message for LLM to retry or adjust.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 30 | 2026-07-28+ | v2 |
Input parameter 'query' lacks constraints. No min/max length, no format guidance, no examples of valid queries. LLMs may pass empty strings or nonsensical input.