OpenGrok MCP provides three read-only tools with clear verb-noun naming (search, list_projects, get_file). Tool descriptions are present and adequate (50-100 chars). Input schemas are visible with type definitions and parameter descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions lack detail on constraints (e.g., maxResults range, file path format). Error handling is minimal; no recovery guidance or actionable error messages. The search tool's complex filtering (full/def/symbol/path/type) lacks mutual-exclusivity documentation. Overall solid foundation but missing output documentation and error guidance expected of production tools.

• Document output schemas for all three tools. Add to tool descriptions: 'Returns: {projects: string[]}' for list_projects, '{file: string, lines: {lineNumber: string, content: string}[]}' for get_file, etc.
• Add constraint details to search() description: 'maxResults: 1 - 500 (default 20). Specify exactly one of: full, def, symbol. Mutually exclusive.'
• Implement error recovery hints. Catch HTTP 404/403/timeout and return: 'Project not found. Call list_projects() to see available projects.' or 'OpenGrok unreachable. Check OPENGROK_URL environment variable.'
• Document get_file() line indexing: 'startLine and endLine are 1-indexed. If omitted, returns entire file. If startLine > file length, returns empty.'
• Add pagination to list_projects(). Accept optional limit (default 50, max 500) and offset parameters. Return {projects: [...], total: N, hasMore: boolean}.