Exploring support for tool calls / MCP server integrations

[KentonParton]

Hey @gabrielste1n first off, thanks for creating OpenWispr. I’ve been using it for voice-to-text and it’s been really useful.

I wanted to ask whether you’d be open to a PR adding support for tool calls and/or MCP server integrations. I built a much shittier version of this myself, and some of the workflows I use often are voice-triggered tools that open frequently visited websites, launch applications, change songs in Spotify, switch windows, and perform other desktop actions.

The broader idea would be to let users define approved tools that an LLM could invoke from voice input, allowing OpenWispr to support not just transcription, but voice-driven actions on behalf of the user.

I’m curious whether this is something you’ve already thought about, whether it fits your vision for the project, and how you might imagine it being designed if support were added.

I’d be interested in contributing an initial implementation or proposal if this is aligned with the direction you’d want to take.