Loading ToolBench
LODESTONE MCP server — lets AI agents query UK Rates & Thresholds directly. Speaks Model Context Protocol over stdio (JSON-RPC 2.0).
Three tools with basic verb-noun naming (list_, get_, search_) and short descriptions (34-72 chars). All have input schemas with proper types. However, parameter descriptions are completely absent, LLMs cannot infer what 'id' or 'query' mean without context. No output schema documentation. No error handling guidance. No pagination support despite search_metrics returning up to 15 results. Tool descriptions lack WHEN/WHY context and dependency hints. This is a minimal, functional implementation but falls well short of production quality.
Get one UK official figure by id — current value, validity dates, official source URL, history.
List all UK official figures available (id, name, category, current value).
Search UK official figures by keyword (e.g. 'minimum wage', 'ISA', 'price cap').
No parameter descriptions. 'id' and 'query' parameters lack any explanation of format, constraints, or examples. LLMs cannot infer whether 'id' is a slug like 'minimum-wage' or an opaque UUID.
No output schema documentation. Tool descriptions do not specify what fields are returned (e.g., does get_metric return 'validity_dates', 'source_url', 'history'?). LLMs must guess the response structure.
search_metrics returns up to 15 results with no pagination support. No limit parameter, no offset/cursor, no total count. Large result sets will waste tokens; no way to fetch additional matches.
Error handling is minimal. get_metric returns a bare error message ('unknown id') with no recovery guidance. No suggestion to call list_metrics or search_metrics first.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2026-07-28+ | v2 |
Tool descriptions lack WHEN/WHY context. 'List all UK official figures' does not explain when to call this vs search_metrics, or that it returns only summary fields (id, name, category, value) not full details.