GoogleDocsMCP has 2 tools with basic verb-noun naming (create_document, update_document) but critical gaps in schema completeness, parameter descriptions, and error handling. Both tools lack output schema documentation. Parameter descriptions are minimal (10-20 chars). No error guidance, no validation, no security considerations. Descriptions are present but generic. This server falls well below production baseline.
Create a new Google Doc and return its document ID.
Update a Google Doc with the given content.
No output schema documented for either tool. create_document returns a string (document_id) but LLMs cannot plan downstream calls without knowing the structure. update_document returns nothing explicitly, forcing agents to guess what happened.
Parameter descriptions are trivial (10-20 chars). 'The title of the document to create' and 'The ID of the document to update' lack context about format, constraints, or examples. Rubric baseline: avg param description 72 chars; these are 30-40 chars.
No error handling or recovery guidance. If document creation fails (auth error, quota exceeded, invalid title), the tool returns nothing. LLMs cannot distinguish retryable from fatal errors or know what to do next.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 37 | 2026-07-28+ | v2 |
Credentials hardcoded in source (SERVICE_ACCOUNT_FILE path). Violates secret-injection pattern, service account keys must never appear in code or tool parameters. Should use environment variables or vault.
No input validation. update_document accepts any string as document_id and content with no length limits, format checks, or sanitization. LLMs can pass malformed IDs or massive content payloads.