MCP server for searching and reading Apache Spark documentation
The server provides 2 tools with clear action verbs (search_, read_) and documented input schemas. Both tool descriptions are present and moderately detailed (85-110 chars), exceeding the 20-char minimum. However, parameter descriptions vary in depth, some lack explicit constraints on valid ranges or formats. The output schema is NOT explicitly documented in the visible code, which is a critical gap. Error handling is present but generic. The server demonstrates competent baseline quality but lacks the depth expected of A-grade tools.
Read the full content of a specific Spark documentation page.
Search Apache Spark documentation by keyword query.
Output schema not documented in tool definitions or descriptions. LLMs cannot plan downstream composition without knowing the response structure (e.g., what fields are returned by search_documentation?).
Parameter 'section' in search_documentation lacks enum constraint. Description lists examples ('sql-ref', 'api', 'streaming', etc.) but does not declare them as an enum, inviting hallucinated values.
Parameter 'limit' in search_documentation declares max: 50, but validation logic is internal (_impl function). LLM has no upfront knowledge that values >50 will be silently capped.
read_documentation accepts a 'path' parameter but provides no guidance on valid formats, character restrictions, or path traversal prevention. Description is minimal (39 chars).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 61 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 56 | - | v1 |
No documented idempotency guarantees. Are repeated calls with the same query safe? This affects retry behavior and multi-step agent plans.