agents-voice
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| voice_speakA | Speak a short, natural voice summary (1-2 sentences) directly to the user's ears. Use this tool at the end of substantive turns to tell the user what was done, without reading code, diffs, or markdown formatting. Keep it concise, natural, and friendly. Args: message: The concise spoken text (e.g. 'Hab den Bug gefixt und Tests sind grün.'). provider: Optional override ('openrouter', 'openai', 'gemini', 'elevenlabs', 'azure', 'custom'). model: Optional model override (e.g. 'openai/gpt-4o-mini-tts', 'elevenlabs/eleven-v4-turbo'). voice: Optional voice override (e.g. 'marin', 'cedar', 'Kore', 'de-DE-Seraphina:DragonHDLatestNeural'). |
| voice_stopA | Immediately stops any currently playing speech audio if the user requests silence. |
| voice_statusA | Checks the status of the voice output system, active provider, configured keys, and bridge. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 3 tools
Each tool maps to a distinct action on the voice output system: producing speech, halting it, and inspecting system state. There is no plausible way to confuse speak with stop or status, and the descriptions reinforce the boundaries.
All three tools use the same snake_case `voice_` prefix followed by a concise action/state noun (speak, stop, status). The pattern is entirely predictable and readable.
Three tools is a tight, well-scoped surface for a narrow domain (TTS output control). Every tool earns its place: emit audio, cancel audio, check readiness — nothing is redundant or missing at this granularity.
The core lifecycle (play, stop, inspect status) is covered, which handles the main agent workflows including silence-on-request. Minor gaps exist for peripheral needs like listing available providers/voices or queueing/pausing, but these are workable around via the speak overrides.