Skip to main content
One command. You’ll have an AI agent running in your project in under 2 minutes.
Polpo TUI
  1. Run Polpo
  2. Setup wizard If no config exists, Polpo opens the setup wizard automatically. It walks you through:
    • Detect providers — scans your env for API keys and OAuth profiles
    • Pick a model — shows recommended models from your detected providers
    • Name your team — creates your first agent with a role
    You need at least one API key (ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY, etc.) or an OAuth login (polpo auth login). See Providers & Models for 22+ supported providers.
  3. Start working You’re in the dashboard. Type what you need — “create a REST API”, “fix the bug in auth.ts”, “write tests for the user module” — hit Enter, and the agent does it.
That’s it. One command, you’re live.

​
Scale up

When you’re ready for more, Polpo orchestrates full teams of agents working in parallel with dependencies and automated quality checks:

​
What happens under the hood

  1. Mission — AI generates tasks with dependencies and agent assignments
  2. Assign — The orchestrator assigns each task to the right agent, respecting dependencies
  3. Execute — Detached runner processes spawn agents with core tools (read, write, edit, bash, glob, grep, ls)
  4. Assess — Three LLM judges score each result on correctness, completeness, code quality, and edge cases
  5. Done — Tasks that pass move to done. Tasks that fail get feedback and retry automatically

​
What’s next

Build Your Team

Add specialist agents, configure roles and constraints.

Your First Mission

Multi-task missions with dependencies and automated assessment.

Talk from Anywhere

Telegram, WhatsApp, Slack, email — talk to Polpo with the apps you already use.

Configuration

Fine-tune retries, quality thresholds, notifications, and provider settings.