Loading ToolBench
Model Context Protocol (MCP) server for generating, executing, and compiling interactive presentations using Manim Community and Manim-Slides
Single tool 'hello_world' with basic schema and description. Tool has a proper input schema with type and description for the 'name' parameter (default='World'). Description is present (67 chars) but generic and does not follow LLM-optimized patterns. No output schema documented. No error handling guidance. Tool naming lacks action verb clarity (should be 'greet_user' or similar). Server declares resources (status, revealjs_config, slides_list) but these are read-only and not tools, they do not contribute to tool quality scoring. Overall definition quality is minimal; the single tool barely meets baseline requirements.
A basic Hello World tool to verify client-server communication.
Tool naming lacks action verb. 'hello_world' does not start with a clear action verb (get_, create_, send_, etc.). Should be 'greet_user' or 'verify_connection' to signal intent to LLMs.
No output schema documented. Tool returns a string but LLMs cannot infer structure or downstream chaining. Document return type and fields explicitly.
Description is generic and does not explain WHEN to use this tool or what it enables. Current description (67 chars) lacks context for LLM selection. Should explain: 'Verifies MCP server connectivity and readiness. Use this first to confirm the server is operational before calling other tools.'
No error handling guidance. Exception handler returns a generic error string. Should categorize errors (retryable vs fatal) and guide LLM recovery.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
Server implements resources (status, revealjs_config, slides_list) but no actual Manim-Slides tools (render, compile, export). Tool set is incomplete for the stated domain. Only a placeholder 'hello_world' exists.