This guide walks you through creating a multi-task mission, executing it, and understanding the results. It assumes you’ve already set up your project and configured your team.
Create a mission
Use AI to generate a mission from a prompt:
polpo mission create "Build a blog API: first create a SQLite database layer with users and posts tables, then build Express REST endpoints on top of it"
polpo mission list
polpo mission show <mission-id>
Task 1: Create database layer
assignTo: backend-dev
dependsOn: []
Task 2: Build REST endpoints
assignTo: backend-dev
dependsOn: ["Create database layer"]
You can also create missions in the TUI with /plan, which offers AI generation and manual editing.
Run the mission
You’ll see output like:
[14:00:01] Polpo started — my-project
[14:00:01] Team agents: backend-dev, frontend-dev
[14:00:02] [task-1] Task added: Create database layer
[14:00:02] [task-2] Task added: Build REST endpoints
[14:00:02] [task-1] Spawning "backend-dev" for: Create database layer
[14:24:15] [task-1] PASSED (score: 4.2/5)
[14:24:15] [task-1] DONE — Create database layer
[14:24:17] [task-2] Spawning "backend-dev" for: Build REST endpoints
[14:25:30] [task-2] PASSED (score: 4.0/5)
[14:25:30] [task-2] DONE — Build REST endpoints
All 2 tasks completed! (1m29s)
Monitor progress
In a second terminal:
This shows a live dashboard with task status, agent activity, and scores. Or start the HTTP server for the Web UI:
polpo serve
# API at http://localhost:3000/api/v1/
Assessment scores
Each task gets scored on 4 dimensions (1-5):
| Dimension | Weight | What It Measures |
|---|
| Correctness | 35% | Does the code work correctly? |
| Completeness | 30% | Are all requirements addressed? |
| Code Quality | 20% | Is the code clean and maintainable? |
| Edge Cases | 15% | Are error conditions handled? |
defaultQualityThreshold).
What if a task fails?
Polpo handles failures automatically:
- Sends detailed feedback to the agent (per-dimension scores + reasoning)
- Enters fix phase — agent makes targeted corrections
- Re-assesses up to
maxFixAttempts (default: 2) times
- Falls back to full retry if fixes aren’t enough
- Gives up after
maxRetries (default: 3) total attempts
Mission task fields
Each task in a mission uses these fields:
| Field | Type | Description |
|---|
title | string | Short task title (must be unique within the mission; used for dependency references) |
description | string | Detailed instructions for the agent |
assignTo | string | Agent name from your team |
dependsOn | string[] | Titles of tasks that must finish first |
expectations | array | Assessment criteria (optional) |
maxRetries | number | Override default retry limit (optional) |
Next steps
