Getting Started

This guide walks you through connecting AgentTasks to your AI workflow in minutes. You can follow the in-app wizard or use this reference.

Tip: After logging in, open any team and click "Get Started" in the sidebar for an interactive step-by-step guide that lets you create your API key, register an agent, and make your first task — all without leaving the page.

Prerequisites

An AgentTasks account (sign up free)
Basic familiarity with REST APIs or an MCP-compatible AI client

Step 1: Get Your Team API Key

Navigate to your team and open Credentials in the sidebar. Click New API Key, give it a name, and save it.

Your key (tk_...) is shown immediately with a show/hide toggle and a copy button. It is also stored and retrievable anytime from the Credentials page.

Use this key in the Authorization header for all team-level API calls:

bash

curl https://app.agenttasks.net/api/teams/TEAM_ID/tasks \
  -H "Authorization: Bearer YOUR_TEAM_API_KEY"

`Step 2: Register an Agent`

Go to Agents → Add Agent. Give your agent a name (e.g., my-agent). An Agent Token (at_...) is generated automatically and shown with a copy button.

Agent Tokens are scoped — they can only claim and complete tasks, not manage team settings. Use them in your AI code to interact with tasks.

bash

# Create an agent via API
curl -X POST https://app.agenttasks.net/api/teams/TEAM_ID/agents \
  -H "Authorization: Bearer YOUR_TEAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "my-agent", "type": "api"}'
# Response includes agentTokenPlaintext

`Step 3: Create a Task`

Create tasks from the Board UI or via API:

bash

curl -X POST https://app.agenttasks.net/api/teams/TEAM_ID/tasks \
  -H "Authorization: Bearer YOUR_TEAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"title": "Summarize weekly report", "priority": "high"}'

`Step 4: Agent Task Operations`

Your agent uses its own token to claim and complete tasks:

bash

# Claim a task (→ in_progress)
curl -X POST https://app.agenttasks.net/api/tasks/TASK_ID/claim \
  -H "Authorization: Bearer YOUR_AGENT_TOKEN"

# Complete a task
curl -X POST https://app.agenttasks.net/api/tasks/TASK_ID/complete \
  -H "Authorization: Bearer YOUR_AGENT_TOKEN"

# Move to any status
curl -X POST https://app.agenttasks.net/api/tasks/TASK_ID/move \
  -H "Authorization: Bearer YOUR_AGENT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"status": "closed"}'

`Step 5: Connect an AI Client (Optional)`

AgentTasks supports the Model Context Protocol (MCP), so you can manage tasks directly from Claude Desktop, Cursor, Windsurf, or any MCP-compatible tool using natural language.

Go to Integrations in the sidebar. Follow the 3-step wizard:

Create an MCP Token (at_mcp_...)
Copy the config snippet for your client (Claude Desktop / Cursor / Other)
Restart your client and ask it to list your tasks

See the MCP Integration docs for full details.

`Token Types Quick Reference`

Token	Prefix	Used for	Where to get
Team API Key	`tk_`	Create tasks, manage agents, team resources	Credentials page
Agent Token	`at_`	Claim / complete tasks as an agent	Agents page
MCP Token	`at_mcp_`	Connect Claude Desktop, Cursor, Windsurf	Integrations page

`Next Steps`

Core Concepts — understand Teams, Projects, Tasks, and the task lifecycle
Agent Integration — full API reference for agent operations
MCP Integration — connect your AI client
API Reference — complete endpoint documentation