FAQ
Questions, answered.
If something's still unclear, drop us an email.
Do I need an API key?+
No. You can run Loclo entirely against a local Ollama model with zero API keys. If you want to use Claude, OpenAI, or another hosted provider, drop your own key in the picker — there's no Loclo-side billing.
Does it work without internet?+
Yes, when paired with a local model like Ollama. The IDE, agent runner, preview server, and verification pipeline are all local. Internet is only required if you're calling a hosted model.
Which models are supported?+
Anything that speaks the OpenAI or Anthropic API shape — Claude (direct API and Claude Code), OpenAI, OpenCode, Codex, Ollama, and any compatible self-hosted endpoint. Switch mid-conversation.
Can I use it on a private or closed-source codebase?+
Yes — that's the point. Loclo runs locally. With a local model, no source code ever leaves your machine. With a hosted model, only the context you send to that provider does.
How is this different from Cursor, Lovable, or Claude Code?+
Cursor is an editor with AI bolted on; Loclo is agent-first. Lovable is single-repo and cloud-only; Loclo handles multi-repo projects on your machine. Claude Code is a CLI; Loclo wraps it (and others) with preview, DevTools, verification, and agent graphs.
What languages and frameworks?+
Anything in your editor and your runner. There's no per-language plugin model — the agent reads files and runs your project's own scripts. TypeScript, Python, Rust, Go, Ruby, PHP — same flow.
How do I update?+
Updates are delivered automatically when connected. If running locally, pull the latest release and restart. Your projects, sessions, and skills persist.
Is my code sent anywhere?+
Only to the model provider you configure, and only the context you ask the agent to consider. No telemetry, no background sync, no Loclo-hosted servers in the loop. Pair with Ollama for zero outbound traffic.