Create a free OpenRouter account, connect it to this minicode serve session, and use free models without editing config files first.
Choose a preset like LM Studio, OpenAI, Ollama, or Custom, then adjust the endpoint if needed before connecting this minicode serve session.
Paste an Anthropic API key to connect this minicode serve session without editing config files first.
minicode needs a model provider to run. Configure one of the following:
Start your local server, then set in ~/.minicode/.env:
MODEL_PROVIDER=openai-compatible OPENAI_BASE_URL=http://localhost:1234/v1 MODEL=your-model-name
Set in ~/.minicode/.env or your shell environment:
MODEL_PROVIDER=anthropic ANTHROPIC_API_KEY=sk-ant-... MODEL=claude-sonnet-4-20250514
Set it manually in ~/.minicode/.env or your shell environment:
MODEL_PROVIDER=openai-compatible OPENAI_BASE_URL=https://openrouter.ai/api/v1 OPENROUTER_API_KEY=sk-or-... MODEL=your-model-name
After updating your config, restart minicode serve.
Persist non-secret global defaults in ~/.minicode/.env for future sessions.
~/.minicode/.env.
Secrets such as API keys stay env-only for now. Saved changes update ~/.minicode/.env and apply to new runs after restart.
Authorize this minicode serve session and optionally save the resulting setup for future runs.
minicode will redirect you to OpenRouter so you can sign in and approve access. By default, the resulting API key is used only for this live serve session.
If you opt in below, minicode will update ~/.minicode/.env after sign-in so future runs can reuse this OpenRouter setup.
When enabled, minicode will store OPENROUTER_API_KEY and set MODEL_PROVIDER plus OPENAI_BASE_URL for OpenRouter. Nothing is written unless you check the box.
OpenRouter handles the sign-in. minicode only receives the API key after you approve access.
Configure an OpenAI-compatible endpoint for this minicode serve session and optionally save it for future runs.
Pick a preset to prefill a common endpoint, then adjust it if needed. minicode will point this serve session at the OpenAI-compatible API endpoint you provide.
Local runtimes like LM Studio or Ollama usually work without an API key, while hosted providers like OpenAI typically require one. The endpoint stays editable no matter which preset you choose.
LM Studio pre-fills the default local server endpoint at http://localhost:1234/v1.
When enabled, minicode will update MODEL_PROVIDER and OPENAI_BASE_URL, and it will also write OPENAI_API_KEY if you provide one. Nothing is written unless you check the box.
This uses the same OpenAI-compatible runtime path minicode already uses for local and hosted providers.
Configure Anthropic for this minicode serve session and optionally save it for future runs.
Paste an Anthropic API key to point this serve session at the Anthropic runtime path. By default, the key is held only by this live session.
minicode will use MODEL_PROVIDER=anthropic for this session. Select an Anthropic model afterward if one is not already configured.
When enabled, minicode will write ANTHROPIC_API_KEY and set MODEL_PROVIDER=anthropic. Nothing is written unless you check the box.
Session-only credentials are cleared when this serve process exits or you disconnect.
Loading…
Loading...
The agent wants to run a tool that modifies your workspace.
To skip future prompts for this session, change the Auto-allow dropdown below the chat input.