An MCP server that provides tools to search and read files from Google Drive
This server has only 2 tools with visible schemas and descriptions. Both tools are read-only operations on Google Drive. Naming follows verb_noun convention (search, read). Schemas are present and properly structured with required fields. However, descriptions are extremely brief (under 50 chars) and lack context about when to use each tool, what they return, or dependencies. Parameter descriptions are minimal. No output schema is documented. Error handling is present but generic. The tool definitions are explicitly visible in the source code (server.py, list_tools method), so they are not inferred.
Read file with FileName in Google Drive
Searches for file with FileName in Google Drive
Tool descriptions are extremely brief (under 50 characters) and lack actionable context. 'Searches for file with FileName in Google Drive' does not explain when to use search vs read, what fields are returned, or how to handle multiple results.
No output schema is documented. The code returns TextContent objects, but the LLM has no formal specification of what fields it receives or what structure to expect from responses.
Parameter descriptions are minimal. The 'fileName' parameter description is just 'FileName to search', it does not explain the search scope, whether partial matches work, case sensitivity, or file type restrictions.
Error handling returns dict objects with free-text 'response' fields (e.g., {'response': 'File not present'}) instead of structured error responses with actionable recovery guidance. LLMs cannot parse free-text errors reliably.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 53 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 40 | - | v1 |
The 'read' tool searches files by name and returns only the first match (if exactly one exists). If the search returns multiple files, it fails with 'More than one file present'. The LLM has no guidance on how to disambiguate, should it search again with a more specific name? Request the user to clarify?
No pagination support. The get_files function hardcodes page_size=10, but the tool interface does not expose limit or offset parameters. Large result sets are truncated silently, and the LLM cannot request more results.
The 'search' tool returns a concatenated string of filenames ('file1.txt, file2.txt, ...') instead of structured objects. LLMs cannot extract individual file IDs, MIME types, or timestamps for downstream operations.