termem has 8 tools with basic schemas and descriptions, but significant gaps limit production readiness. Tool names lack consistent verb prefixes (e.g., 'recall_sessions' vs 'list_sessions', 'interactive_picker' is vague). Descriptions are present but brief (avg ~60 chars, below the 194-char baseline). Parameter descriptions exist but lack detail on constraints, formats, and dependencies. No output schemas documented. Error handling and recovery guidance absent. No tool annotations (readOnlyHint/destructiveHint). The server is STDIO-only, which is a hard transport limitation.

• Rename tools to follow verb_noun convention: 'recall_sessions' → 'list_sessions', 'interactive_picker' → 'pick_session', 'mcp_server' → 'start_mcp_server'. This makes intent clear from the name alone.
• Document output schemas for all tools. E.g., recall_sessions should return {sessions: [{id, title, cwd, source, created_at}], total_count, limit}. This enables LLMs to chain tools without discovery calls.
• Add enum constraints to parameters. E.g., source: enum=['claude', 'codex', 'shell']; target: enum=['zsh', 'bash', 'powershell', 'claude', 'codex', 'opencode', 'gemini', 'pi']. Replace comma-separated text with formal constraints.
• Add min/max bounds to numeric parameters. E.g., limit: {type: integer, minimum: 1, maximum: 1000}. This prevents LLMs from passing absurd values.
• Add error handling guidance to tool descriptions. E.g., resume_session: 'If no session matches the query, returns error with available sessions in the directory. Call list_sessions to see options.' This guides agent recovery.