FIELD GUIDE / PROVIDERS
Bring the model
you want to work with.
Veasel Code keeps model connections behind a small V provider interface. Configure one backend and use the same persistent session workflow across OpenAI-compatible APIs, Anthropic, and Gemini.
Available in the current source build
The V backend sends bounded synchronous completions, keeps credentials server-side, and stores each successful user/assistant exchange in the session database. Streaming, tool use, and repository edits are still being built.
Choose a provider for the backend
export VEASEL_MODEL_PROVIDER=anthropic
export VEASEL_MODEL=your-model-name
export ANTHROPIC_API_KEY=your-api-key
v run . serveUse OPENAI_API_KEY withopenai-compatible, or GEMINI_API_KEY withgemini. Set VEASEL_MODEL_BASE_URL for a compatible endpoint; custom endpoints require HTTPS or loopback HTTP. The TUI inherits provider configuration from the server process.
Today: configured sessions can chat with a model and restore recent history. Veasel cannot inspect files, apply patches, or run commands yet. See theimplementation planfor current progress.
See the architecture