MCP server providing RAG (Retrieval-Augmented Generation) capabilities for Land Rover Discovery 2 documentation using ChromaDB vector database and Embed4All embeddings
Single tool with basic schema and description. Tool name 'search_knowledge_base' follows verb_noun convention (+). Description is present but generic (58 chars, below 194-char baseline). Input schema visible with types and descriptions for both parameters (+). However, output schema is completely undocumented, function returns a plain string with no structured format specification. No error handling guidance, no pagination despite potentially returning multiple results, no field documentation for chaining. Missing 80% of production-grade patterns.
Search the ChromaDB knowledge base for documents relevant to the user's query.
Output schema completely undocumented. Function returns plain string with no structured format, field names, or type information. LLMs cannot parse results reliably or extract data for downstream tool calls.
Tool description is generic and under 100 chars (58 chars). Does not explain WHEN to use this tool, what the output structure is, or how results should be interpreted. Baseline is 194 chars.
No pagination support despite returning multiple results (n_results parameter). Large result sets will blow context windows. No limit enforcement, no next_cursor, no total_count in response.
No error handling guidance. If ChromaDB query fails, embedding fails, or collection is missing, function will raise unhandled exceptions. No recovery hints for LLM.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
Parameter 'n_results' lacks constraints. No minimum/maximum specified. LLM could pass 0, negative, or 10000+ values, causing API errors or resource exhaustion.