Loading ToolBench
Codebase registry — reference patterns across your projects. MCP server that manages a registry of projects and their patterns, with analysis and search capabilities.
Server defines 6 tools with clear verb-noun naming (list_, search_, get_, read_). All tools have descriptions (avg 95 chars, within baseline 34-392). Input schemas present for all tools with type declarations. However, parameter descriptions are sparse or missing in several tools (e.g., list_nodes lacks layer description detail), and output schemas are not documented. Error handling is minimal, no recovery guidance or categorization. Tool composition is sound (single responsibility), but descriptions could be more LLM-optimized per pattern:tool-description baseline.
Get full details for a specific registered project including all patterns, links, and metadata. Use the project name from list_projects.
List nodes in the analyzed project graph.
List all registered projects in the codebase registry with their descriptions, tags, and patterns
Read a specific file from a registered project. Path must be relative to the project root. Supports reading specific line ranges for large files.
Search for a text pattern in a registered project's source code. Returns matching lines with file paths and line numbers. Case-insensitive.
Search registered projects by keyword — matches against names, descriptions, tags, and pattern names/descriptions
list_nodes has minimal description ('List nodes in the analyzed project graph.') and vague parameter documentation. 'layer' parameter lacks type, range, or valid values. Does not explain when to call this vs other discovery tools.
No output schemas documented for any tool. LLMs cannot infer what fields to expect (e.g., does list_projects return 'id' or 'name'? Is it paginated?). Breaks downstream tool chaining and forces agents to guess field names.
search_project_code and read_project_file accept optional parameters (fileGlob, startLine, endLine) but lack guidance on defaults, ranges, or constraints. E.g., what happens if startLine > endLine? What is the max file size?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
No error handling or recovery guidance. Tools return raw errors (e.g., 'file not found', 'project not found') without suggesting next steps. Per pattern:recovery-guide, errors should tell LLMs what to do next.
search_project_code description includes example values ('Upload', 'handleSubmit', 'useState.*modal') which LLMs may reuse literally. Per pattern:tool-description, use formal constraints (regex patterns, enums) instead of examples in descriptions.