Tools
Gemini Stt
By Gemini speech-to-text for OpenClaw
Install
npm install
npmREADME
# Gemini 3.5 Transcribe for OpenClaw
Registers `gemini-3.5-transcribe` as an OpenClaw audio media-understanding provider.
OpenClaw's bundled Google audio provider only calls `:generateContent`, which the
dedicated Transcribe model does not serve; this plugin uses the Google Interactions
API instead.
## Auth
`GEMINI_API_KEY` (or `GOOGLE_API_KEY`) in the Gateway environment, or
`models.providers.gemini-transcribe.apiKey`.
## Config
```json5
{
tools: {
media: {
models: [
{ provider: "gemini-transcribe", model: "gemini-3.5-transcribe", capabilities: ["audio"] },
{ type: "cli", command: "whisper", args: ["--model", "base", "{{AttachmentPath}}"], capabilities: ["audio"] }
],
audio: { enabled: true, preferredModel: "gemini-transcribe/gemini-3.5-transcribe" }
}
}
}
```
## Development
```bash
npm install
npm run build # tsc -> dist/index.js
npm test # vitest (registration, auth, request shape, errors)
```
Install locally with `openclaw plugins install . --force`. Provider plugins are
validated at runtime by `openclaw plugins doctor`; `openclaw plugins validate`
and `plugins build` only cover tool/feature authoring metadata and intentionally
do not apply to media-understanding providers.
tools
Comments
Sign in to leave a comment