Providers
Configure any OpenAI-compatible endpoint. Multiple profiles are supported — one is active at a time. Values are stored only in this browser.
New profile
Saved profiles (0)
No profiles yet. Add one above.
Local agents
Detects OpenAI-compatible servers on well-known localhost ports (Ollama, LM Studio, opencode, Claude Code, codex, gemini, hermes). Detection requires the agent to expose an HTTP endpoint; some CLIs do not — nothing to import in that case.
GitHub token — optional
Anonymous requests to GitHub are limited to 60/hour. Adding a fine-grained personal token raises this to 5000/hour. Public-repo read is enough — no write scopes needed.