Four tools with complete JSON schemas and descriptions. Naming follows verb_noun pattern (browse_, get_, mutate_, delete_). Descriptions are adequate (50-200 chars) but lack LLM-optimized guidance on when to use each tool vs. alternatives. Parameters are well-typed with enums and constraints (e.g., transcript_block enum, limit bounds). Critical gap: no output schemas documented, LLMs cannot predict response structure for planning. Error handling and recovery guidance absent. Destructive tool (delete_recording) requires explicit confirmation, which is good, but no dry-run or undo patterns for mutate_recording.

• Document output schemas for all tools. Example for browse_recordings: {recordings: [{id, title, created_at, folder_id, duration_seconds}], next_after: integer|null, total_count: integer}. This enables LLMs to chain tools and extract IDs for follow-up calls.
• Split mutate_recording into four single-action tools: rename_recording(recording_id, new_name), trash_recording(recording_ids: []), restore_recording(recording_ids: []), move_recording(recording_ids: [], folder_id). Each tool has one clear responsibility and reduces parameter confusion.
• Add a list_folders() tool to discover folder IDs. Currently mutate_recording references folders but no discovery mechanism exists, forcing users to know folder IDs out-of-band.
• Enhance parameter descriptions with format constraints: 'since: ISO 8601 datetime (UTC, inclusive); e.g., 2024-01-15T09:00:00Z'. Add timezone and boundary semantics.
• Add error recovery hints to descriptions. E.g., delete_recording: 'Permanently delete a recording. This cannot be undone. If you need to recover, restore from trash first using trash_recording() and restore_recording().' This guides LLMs toward safer patterns.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry