Loading ToolBench
Multi-project RAG + MCP server for coding agents. Provides document retrieval, search, and context extraction across multiple Git repositories with vector embeddings.
RAGBuddy defines 4 tools with basic descriptions and input schemas. All tools follow verb_noun naming (get_*, search_*, list_) and have non-empty descriptions (range 80 - 180 chars, within baseline 34 - 392). However, parameter descriptions are minimal or absent, output schemas are undocumented, and error handling lacks recovery guidance. Tools are READ_ONLY and well-scoped, but lack the depth expected for production use. No tool annotations (readOnlyHint, etc.) despite all being read-only. Schemas are present but sparse, parameters lack format/constraint details.
Retrieves comprehensive context for a project including vector-based semantic search results and metadata
Retrieves the full content of a specific project document by file path
Lists all indexed documents and chunks for a project with metadata
Searches project documentation using vector embeddings and BM25 hybrid search
Output schemas undocumented. No visible return type definitions for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., does search_project_docs return chunk_id, score, file_path?).
Parameter descriptions missing or trivial. 'project' param described only as 'Project ID to retrieve context for', no format, length, or example. 'query' param lacks guidance on expected length, language, or semantic scope.
No error handling guidance. Tool definitions lack recovery hints. If search_project_docs returns no results, should the LLM retry with a different query, broaden the project scope, or ask the user? No guidance provided.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2026-07-28+ | v2 |
Tool annotations missing. All 4 tools are read-only (risk: READ_ONLY), but no readOnlyHint annotation in schema. Agents cannot infer safety from tool metadata.
No pagination or result limits documented. list_project_knowledge and search_project_docs may return unbounded results. No limit, offset, or cursor parameters visible. Large result sets will exhaust context windows.