Install, connect models, define wallet walls, then pick the operating mode.
- API server defaults to port 3377
- OAuth and API-key providers can coexist
- Risk limits remain operator-defined and non-negotiable
SETUP
WhiteOwl is designed for local-first operation. Installation, provider routing, wallet risk, and runtime mode selection are separate steps so the operator can harden each one deliberately.
BOOT FLOW
git clone https://github.com/yourname/WhiteOwl.git
cd WhiteOwl
npm install
npx tsx src/index.ts server
The API and dashboard default to port 3377. Keep the runtime local and use the dashboard as the control plane.
OPENAI_API_KEY=...
ANTHROPIC_API_KEY=...
OAUTH_GOOGLE_CLIENT_ID=...
OAUTH_AZURE_CLIENT_ID=...
OLLAMA_BASE_URL=http://localhost:11434
OAuth-backed GitHub, Google, and Azure flows can coexist with direct API-key providers and local Ollama.
SOLANA_RPC_URL=...
WALLET_PRIVATE_KEY=...
MAX_POSITION_SOL=0.5
MAX_DAILY_LOSS_SOL=2
MAX_OPEN_POSITIONS=4
Risk walls stay operator-defined. The AI layer can work inside them, but not around them.
npm run monitor
npm run autopilot
npm run report
Use monitor for read-only flow, autopilot for autonomous execution, and report mode for recent outcome review.
CHECKLIST
At least one fast model, one strategic model path, and a fallback route should be configured before sessions go live.
RPC, signing material, and position caps need to be aligned with the actual risk budget of the operator.
Dashboard, extension, and alert channels should be reachable before autonomous flow is allowed to continue unsupervised.
MODEL ACCESS
WhiteOwl can route across OAuth-backed and API-key-backed providers without changing the operator surface.
READY
Use the runtime page for deeper system logic and the extension page for the browser-side wallet and inspection stack.