MCP server for ElevenLabs Text-to-Speech integration with OpenCode
Single tool with basic schema and description. Tool name 'text_to_speech' is action-oriented and clear. Description is adequate (67 chars) but lacks context on when to use it or what it returns. Input schema is present with proper types and defaults, but output schema is completely undocumented, LLMs cannot plan downstream operations. Parameters lack detailed constraints (e.g., text length limits, valid voice_id formats). Error handling returns generic messages without recovery guidance. No parameter descriptions explain format expectations or constraints.
Convert text to speech using ElevenLabs API
Output schema completely undocumented. Tool returns text content but LLMs cannot infer structure, fields, or what to do with the result. No documentation of whether audio file path, URL, or metadata is returned.
Parameter descriptions missing or minimal. 'text' has description but 'voice_id' and 'model_id' descriptions are vague (e.g., 'Voice ID (default: CwhRBWXzGAHq8TQ4Fs17 - Roger voice)' does not explain valid format, how to discover voice IDs, or constraints). LLMs cannot validate inputs.
No input validation or constraint documentation. 'text' parameter has no length limits, encoding constraints, or language restrictions documented. LLMs may pass extremely long strings or unsupported languages without guidance.
Error messages lack recovery guidance. Code returns generic 'Error generating text-to-speech: {error.message}' without actionable next steps. LLMs cannot determine if error is retryable, user-fixable, or fatal.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
Tool description does not explain what the tool returns or when to use it. 'Convert text to speech using ElevenLabs API' is functional but lacks context: Does it play audio? Save a file? Return a URL? What format? This forces LLMs to guess.