Loading ToolBench
MCP server that displays the current time
Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. get_current_time has reasonable parameter descriptions but lacks output schema documentation. search_ndl_books has minimal parameter descriptions and no guidance on API response structure. Neither tool documents what fields agents should expect in responses, violating the pattern:tool-chain requirement. Error handling is present but generic. No tool annotations (readOnlyHint, etc.) despite both being read-only operations.
Get the current time in various formats
Search books in the National Diet Library using keywords
Output schemas not documented. LLMs cannot plan downstream tool calls or extract structured data from responses. get_current_time returns JSON with 'current_time', 'format', 'timezone' fields, but this is not declared in tool definition. search_ndl_books returns raw XML/text from NDL API with no schema guidance.
search_ndl_books parameter descriptions are minimal. 'query' lacks format guidance (free-form text? boolean operators?). 'maxResults' has no range constraints (1-100? 1-1000?). Rubric baseline: 100% of A+ tools have param descriptions; this tool has only 1 of 2 params adequately described.
No tool annotations despite both tools being read-only. Current spec (2026-07-28) supports readOnlyHint in tool definitions. Both tools should declare readOnlyHint: true to signal to agents that they are safe to call without side effects.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | 2026-07-28+ | v2 |
get_current_time description lacks WHEN guidance. 'Get the current time in various formats' does not explain when to use this vs other time tools or what the timezone parameter actually does (does it convert, or just format?). Rubric: descriptions should state WHAT, WHEN, and prerequisites.
search_ndl_books returns raw XML text without parsing. Response is unstructured, forcing LLMs to parse XML on the fly. Rubric baseline: 100% of A+ tools return structured objects with typed fields. Should parse NDL response and return array of {title, author, url, publication_date} objects.