Parchi is a sidebar agent that reads pages, fills forms, clicks buttons, and navigates tabs. Just tell it what to do.
Anthropic, OpenAI, OpenRouter, Kimi, or any compatible endpoint. Switch mid-conversation.
Read pages, click elements, fill forms, navigate tabs, take screenshots. All via natural language.
Bring your own API keys for zero cost. Or use Stripe-managed billing for instant, keyless access.
Automatic screenshots analyzed by vision models. Sees exactly what you see.
Save model configs as profiles. Record browser flows as replayable skills.
MIT licensed. Self-host the backend, audit every line, extend with your own tools.