TCalc MCP Server defines 2 tools with moderate quality. Both tools have descriptions and input schemas present, but significant gaps exist in parameter documentation, error handling guidance, and output schema clarity. Tool names follow verb_noun conventions appropriately (create_repo_map, compact_completed_goal). However, parameter descriptions are sparse or missing, and no output schemas are documented. The server implements a workspace analysis domain with READ_ONLY risk classification, but lacks production-grade error recovery patterns and actionable error messages. Most parameters in create_repo_map have descriptions; compact_completed_goal has mixed parameter coverage. No examples of error handling patterns that guide LLM recovery are evident in the provided source excerpts.

• Document output schemas for both tools. For create_repo_map, define: { type: object, properties: { repositoryMap: { type: string, description: '...' }, tokenCount: { type: number }, ... } }. For compact_completed_goal, define: { type: object, properties: { compactedSummary: { type: string }, status: { type: string, enum: [...] }, ... } }.
• Enhance compact_completed_goal parameter descriptions. Add descriptions for 'sourceTokens' (e.g., 'Token count of the source files involved in the completed goal, used for optimization metrics'). Clarify 'goal' enum selection: 'The type of workspace goal that was completed; used to categorize the compaction and inform recommendations for follow-up work.'
• Add recovery guidance to tool descriptions. E.g., for create_repo_map: 'If tokenBudget is too low for the full map, consider increasing it or disabling enableSymbolExtraction to reduce output size.'
• Document parameter relationships. For create_repo_map, explain: 'The goal parameter filters the repo map to prioritize files relevant to that goal. If not provided, the full repository structure is returned.'
• For enableSymbolExtraction, add detail: 'When true, extracts function names, class definitions, and exported constants. Increases output size; recommended when building agent context for code navigation tasks. Set maxSymbols to limit extraction (e.g., 100 symbols per file).'
• For format parameter, clarify: 'Choose markdown for human-readable summaries; choose json for programmatic parsing and downstream tool integration.'