Loading ToolBench
The AI-native health data engine — collect, standardize, and reason over labs, wearables & genomics.
mirobody has 5 tools with reasonable health-domain naming (resolve_indicator, query_health_indicators, etc.). All tools have descriptions and basic parameter schemas. However, there are significant gaps: parameter descriptions are sparse or missing, output schemas are entirely undocumented, error handling is absent, and the schema completeness is uneven. The server demonstrates domain understanding but falls short of production-grade tool quality. Tools like query_health_indicators accept 'keywords' and 'start_time' but lack descriptions of expected formats, constraints, or return structure. The example_echo tool is present but contributes minimal value.
Return the text you sent, so you can see a plugin tool round-trip. Use it once to confirm the plugin is installed; it has no other use. `text` is echoed back unchanged.
Query a user's genetic data and genomic analysis results.
Query a user's stored health indicator records (readings, measurements, lab values).
Query a user's stored medication records and medication history.
Resolve indicator names (LOINC, SNOMED, etc.) from human-readable or multilingual text.
NO OUTPUT SCHEMAS DOCUMENTED for any health query tool (resolve_indicator, query_health_indicators, query_medications, query_genetic_data). LLMs cannot know what fields to expect, forcing guessing and context loss between tool calls.
Parameter descriptions are incomplete or missing required format/constraint details. 'start_time' is typed as string across 3 tools but no format specification (ISO 8601? Unix timestamp? Natural language?). LLMs will guess formats and pass invalid values.
NO ERROR HANDLING GUIDANCE. Tools do not document what happens if keywords don't match, time windows are invalid, or no data exists. LLMs receive bare failures with no recovery path.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | <=2025-11-25 | v2 |
Pagination and result limits not documented. Large result sets could exhaust context; no documentation of page size, total count, or next_cursor mechanism.
Generic parameter naming in query tools. 'keywords' is vague, should clarify intent (e.g., 'indicator_keywords', 'medication_names', 'genetic_variant_ids'). Multi-purpose tools conflate concerns.
example_echo tool is a test utility with no clinical/health value. Should not ship in production, adds clutter to tool discovery and wastes LLM reasoning cycles.