Skip to main content
Run Polpo on your machine — API server and agent execution. No account required. There are two ways to run locally:
  • npm packagenpm install -g polpo-ai. Includes the CLI, API server, and orchestrator. Best for most users.
  • From source — clone the repo. Includes everything above plus the web dashboard. Best if you want the full UI or want to contribute.

From npm package

Initialize and start

polpo init
polpo start
polpo init walks you through provider selection, model choice, and creating your first agent. polpo start starts the API server and orchestrator. API at http://localhost:3890/api/v1.

From source

Clone the repository to get the full stack including the web dashboard:
git clone https://github.com/lumea-labs/polpo.git
cd polpo
pnpm i && pnpm dev:all
Dashboard at http://localhost:5173, API at http://localhost:3890/api/v1. The dashboard includes its own setup wizard — follow the on-screen steps to configure your provider, model, and first agent.

Dashboard

From the dashboard you can:
  • Chat with your agents
  • Create and monitor tasks and missions
  • Manage agent configurations, memory, and skills
  • View sessions, logs, and assessment results
  • Configure teams and orchestration settings
The dashboard is only available when running from source. It is not included in the polpo-ai npm package. If you installed via npm, use the cloud dashboard instead.

Using the SDK locally

Point the SDK to your local server:
import { PolpoClient } from "@polpo-ai/sdk";

const client = new PolpoClient({
  baseUrl: "http://localhost:3890",
});

const stream = client.chatCompletionsStream({
  agent: "backend-dev",
  messages: [{ role: "user", content: "Hello" }],
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}
No API key is required for local development by default. Add --api-key <key> to polpo start if you want to protect the local API.

Local to cloud

When you’re ready to deploy:
polpo login
polpo deploy
Same config, same agents. The CLI detects local LLM keys and offers to push them to the cloud. See Deploy for more details.

Options

FlagDescription
--port <port>Port to listen on (default: 3890)
--host <host>Host to bind to (default: 127.0.0.1)
--dir <path>Working directory (default: .)
--api-key <key>Protect the API with a key
--setupLaunch the setup wizard in the dashboard
--cors-origins <origins>Allowed CORS origins (comma-separated)

See also

For production self-hosting (not just local dev), see Self-hosting — covers Docker, reverse proxy, PostgreSQL, and environment variables.