Critical gaps across all dimensions. Tool definitions lack input schemas (cannot verify from source), descriptions are minimal or absent, and parameter documentation is missing. Only 1 of 6 tools has a visible description. No output schemas documented. Error handling and recovery guidance absent. STDIO transport limits remote accessibility. This server falls well below production baseline.

• Add explicit input schemas to all 5 tools lacking them. Each schema must include parameter types (string, number, boolean, array), descriptions (50-100 chars), and constraints (enums, min/max, patterns). Example: list_sessions should document whether it accepts filters, pagination params, and what fields it returns.
• Write descriptions for all 6 tools following the pattern: WHAT does it do? WHEN should the LLM call it? WHAT does it return? Target 50-200 characters. E.g., 'list_providers: Retrieve available transcript providers (Cursor, Claude Code, Codex). Call this first to discover which providers have sessions available.'
• Document output schemas for all tools. Specify returned fields, types, and structure. For list_* tools, include pagination (limit, offset, total_count) and result limits. For search_transcripts, document result ranking and relevance scoring.
• Add parameter descriptions to build_index. Explain what 'full' and 'semantic' indices do, when to rebuild each, and expected duration. Clarify whether both can be true simultaneously and what happens if both are false.
• Implement error handling with recovery guidance. Return structured errors with: error code, human-readable message, constraint violated (if applicable), and suggested next action. E.g., 'Provider not found: got "Codex", available: Cursor, Claude Code, Codex. Call list_providers() to see all options.'