Loading ToolBench
Governed MCP and API gateway for GIS AI GO - provides catalogue search/describe, evidence inspection, and data query operations with policy-based access control
Server has 7 tools with explicit schemas and descriptions. Naming follows verb_noun patterns (search, describe, inspect, resolve, query). Descriptions are present but brief (avg 80 chars, below 194-char baseline). Schemas are well-formed with proper types and enums. Critical gaps: no error handling guidance, no output schemas documented, no parameter validation rules stated, no idempotency hints for read-only operations. Tools 6-7 (explorer_*) duplicate tools 1-2 with weaker schemas (missing enum constraints on facet types). No tool annotations (readOnlyHint present in metadata but not in schema). Composition is sound, tools chain logically (search → describe → resolve → query). Missing: recovery guidance, actionable error messages, output field documentation.
Retrieve detailed information about a specific catalogue record
Search the public catalogue for records matching query and facet filters
Execute a query against selected geospatial data
Inspect evidence receipts and integrity information for catalogue records
Page-scoped catalogue record description for the WebMCP Explorer
Page-scoped catalogue search for the WebMCP Explorer
Resolve a selection of catalogue records for data query operations
No output schemas documented. LLMs cannot plan downstream calls or extract required fields (e.g., what does catalogue.search return? Does it include record_id for chaining to catalogue.describe?). Violates pattern:tool requirement.
Descriptions are too brief (avg 80 chars vs 194-char baseline). Missing WHEN to use each tool and prerequisites. E.g., 'Retrieve detailed information about a specific catalogue record' does not explain when to call this vs search, or what structure it reveals.
Tools 6-7 (explorer_*) duplicate tools 1-2 with degraded schemas. explorer_search_catalogue facets lack enum constraints on 'types' (tools 1 has ['dataset','service','document'], tool 6 has no enum). Creates ambiguity and forces LLM to guess valid values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2025-06-18+ | v2 |
No error handling guidance. If catalogue.search returns 0 results, what should the LLM do? Refine query? Try different facets? No recovery hints provided. Violates pattern:recovery-guide.
Tool annotations (readOnlyHint) present in metadata but not exposed in schema. LLMs cannot see which tools are safe to retry or have side effects. Violates current spec (2026-07-28) requirement for tool annotations in schema.