MCP server for searching and retrieving movie information from a dataset and web
MovieServer has significant definition quality gaps. Only 2 tools are present. Tool naming follows verb_noun convention (get_movie_info, search_web), which is good. However, descriptions are minimal (under 50 chars for both), and critically, the search_web tool has NO description at all in the code. Parameter descriptions are present but sparse. Neither tool documents its return schema, get_movie_info returns a string and search_web returns 'dict | None', neither of which are formally specified. No error handling guidance is provided. The search_web tool lacks explicit input schema documentation in the code (only inferred from function signature). No pagination, no structured output guidance, and no mention of timeout or failure recovery.
Search and return movie info from the dataset.
search_web tool has NO description. LLM cannot determine when or why to select this tool.
get_movie_info description is only 44 chars ('Search and return movie info from the dataset.'), too terse. Does not explain WHEN to use it, what dataset it searches, or how to interpret results.
No output schemas documented. get_movie_info returns a string; search_web returns 'dict | None'. LLMs cannot plan downstream calls or extract structured data.
search_web tool has a hardcoded timeout (30s) but no error handling guidance. If timeout occurs, tool returns {'organic': []}, indistinguishable from an empty result. LLM cannot distinguish between 'no results' and 'service failed'.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 36 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 27 | - | v1 |
search_web accepts an API key via environment variable (SERPER_API_KEY) but does not document this prerequisite or explain what happens if the key is missing.
get_movie_info parameter 'title' has a description ('The movie title to search for') but no mention of case sensitivity, partial matching behavior, or what happens if multiple matches exist (currently returns first row only).
search_web parameter 'query' has a description ('The search query') that is too generic. Does not explain what the query returns, what format to expect, or limits on number of results (hardcoded to 2).
No error classification or recovery guidance. If search_web fails (network error, auth error, malformed response), LLM receives no guidance on whether to retry, ask the user, or abandon the request.