MCP server for playing notification sounds and synthesizing text-to-speech using VOICEVOX on the current machine
Two tools with complete input schemas and descriptions. play_mcp_notification_sound has clear naming and well-structured schema with optional parameters and defaults. speak_text has richer parameter set with numeric constraints (minLength, maxLength, min/max for scales). Both tools include output schemas. However, descriptions are brief (under 100 chars) and lack context about when to use each tool or what happens on failure. Parameter descriptions are present but minimal. No error recovery guidance. Tool composition is reasonable (separate concerns), but descriptions don't explain dependencies or prerequisites.
Play a local notification sound file on the current machine.
Synthesize text with VOICEVOX and play it on the current machine.
Tool descriptions lack actionable context. 'Play a local notification sound file' does not explain when to use this vs speak_text, what happens on file-not-found, or whether playback blocks the agent.
No error recovery guidance in tool descriptions. Errors return structured responses with 'error' and 'details' fields, but descriptions do not explain what errors are possible or how the agent should respond.
Parameter descriptions are minimal. 'soundPath' says 'Optional relative path under sounds/' but does not explain the resolution logic, what happens if the path is invalid, or the expected directory structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | 2025-06-18+ | v2 |
speak_text numeric parameters (speedScale, pitchScale, intonationScale, volumeScale) have min/max constraints in schema but descriptions do not explain the semantic meaning (e.g., 'speedScale 0.5 = half speed, 2.0 = double speed').
No documentation of output schema fields. playSoundResponse and speak response structures are defined in code but not documented in tool descriptions, forcing LLMs to infer field meanings.