Enhances AI capabilities with comprehensive workspace context
This server presents a critical evaluation challenge. While all 6 tools have reasonable naming conventions (verb_noun pattern: analyze_workspace, search_codebase, etc.) and descriptions present in the tool manifest (10 - 134 chars each), the actual IMPLEMENTATION code visible is transpiled JavaScript (.js) rather than TypeScript source (.ts). The source code snippet shows only the compiled output from src/index.js, not the raw TypeScript. This means: (1) input schemas ARE visible and present for all tools, (2) descriptions are visible at tool registration, (3) BUT parameter descriptions within those schemas are minimal or absent in the visible code. For example, 'analyze_workspace' shows parameters 'includeContent' and 'maxDepth' with brief descriptions ('Include file contents in analysis', 'Maximum directory depth to analyze') but no type constraints on maxDepth (numeric range), no validation guidance, and no error recovery hints. All 6 tools follow the READ_ONLY pattern correctly (security-positive), but they operate on workspace analysis, a read-safe domain. The server provides no evidence of pagination, result limiting, output field documentation, or error recovery guidance. Parameter descriptions are present but terse (average ~40 chars), falling below the baseline of ~72 chars for A+ tools. No tool provides examples of output structure or downstream chaining hints. The protocol uses prompts and resources capabilities (positive) but toolAnnotations are absent, which would clarify read-only intent at the protocol level. Most critically: because this is a STDIO-only server, it cannot be used by hosted/remote MCP clients, a hard transport limitation.
Analyze the entire workspace structure and provide comprehensive insights
Get a comprehensive summary of the workspace for AI context
Analyze dependencies and relationships for a specific file
Extract common patterns, architectures, and conventions from the codebase
Search for code patterns, functions, classes, or specific content across the workspace
Analyze the codebase and suggest improvements based on patterns and best practices
STDIO transport only, hard cap of 50 on protocol readiness. This server cannot be used by hosted MCP clients (Claude.ai, Anthropic API with stateless HTTP). Requires local desktop installation only.
Parameter descriptions lack depth and validation guidance. 'Maximum directory depth to analyze' does not specify range (1 - 100? 1 - 1000?), no default behavior if exceeded, no penalty/timeout warning. Baseline is ~72 chars per param; most here are 30 - 50 chars.
No documented output schemas or examples. LLMs cannot plan downstream tool calls (e.g., if analyze_workspace returns file paths, the LLM needs to know the field names and types to feed them to search_codebase or get_file_dependencies). Missing: field descriptions, type info, example response structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 58 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 45 | - | v1 |
No result limiting or pagination. 'search_codebase' with a broad regex could return thousands of matches; no mention of a limit parameter, no max result count in description, no pagination guidance. Large responses bloat context window and degrade LLM reasoning.
No error recovery guidance. If 'get_file_dependencies' fails on a nonexistent file, no actionable error message telling the LLM to try search_codebase first or check the file path. Error handling is absent from tool definitions and likely implementation.
toolAnnotations absent. All 6 tools are READ_ONLY and idempotent; declaring readOnlyHint=true at the protocol level would clarify intent and prevent agents from mistakenly treating them as data-modifying operations.
Tool composition gaps. 'suggest_improvements' accepts only 'scope' enum (security|performance|maintainability|all); there is no 'file' or 'area' parameter to narrow scope. Agents cannot request improvements on a specific file or module, only whole-codebase suggestions, which is coarse-grained.