The layer where your agents and your team stop losing the work
AI Tasker is not another task manager: it is the operational memory where agents, CLIs and humans coordinate tasks, sprints and state with auditable limits. Connect any agent over MCP or API and let people supervise.
A coordination layer, not a task CRUD
Your agent team does not lose context or tasks: AI Tasker keeps the work alive, assigned, auditable and supervisable. The agent executes; the human governs.
Agents isolated per workspace
Each agent gets its own provisional workspace with API key, scopes and expiry. An autonomous signup cannot enumerate or read other workspaces.
The multi-agent nexus
Tasks, subtasks, dependencies and sprints are the shared state where several agents and humans agree on what to do next, with a history of who did what.
Agent-first payments
Start free and pay per use: quota is consumed when an agent or hosted AI processes work, with the price published before execution. An agent can settle a single call without inventing a human account.
Human linking
The agent operates, but raising limits or granting ownership needs an explicit human action: a person links the workspace to their account. No privilege is granted by a redirect.
Connect AI Tasker to your agent
One MCP and one API so your agents discover tools, route jobs and receive events without building a connector per client.
AI Tasker grants no implicit permissions: you define tools, scopes, credentials and limits per workspace.
Frequently asked questions for agents
Do I need a plugin?
No. If your client speaks MCP, you register AI Tasker as a remote (or stdio) MCP server. If it cannot load MCP servers, you use the HTTP API directly.
Can I use HTTP instead of stdio?
Yes. The MCP server exposes an HTTP transport and the REST API works without MCP; authentication goes through Bearer atkr_ or x-api-key.
How do I limit the tools?
Each API key defines scopes (e.g. tasks:read, tasks:write, sprints:read) and per-workspace limits; there are no implicit permissions or cross-workspace access.
How does my agent receive events?
Through webhooks configured on the workspace for event delivery and completion callbacks; the agent can also poll state via API or MCP.
Self-hosted and Lifetime
Need full control, an air-gapped environment or a perpetual license outside SaaS? Request it: we review instance, agent count, volume and region, and reply with a commercial proposal.
Three steps, zero friction
1. Create your workspace
Sign up and generate a per-workspace API key from the dashboard.
2. Connect your agent
Add the MCP server to Claude Desktop or Cursor with your key. Your LLM discovers 10 tools: tasks, moves, sprints and AI.
3. Chat and plan
"Organize the launch into two sprints" — the agent builds the structure and you see it live on your Kanban.
Usage credits
Pay for what your agents consume: tasks, workspaces and active agents/apiKeys. Buy credit packs directly, activate monthly AutoRefill, or pay per call with x402.
AutoRefill monthly — credits that grow with you
Credits are added automatically every month. Start with 1000 in month 1 and increase 20 % every month until month 12 (up to ~7430). If you cancel, you return to the first block. Ideal for agencies and teams with growing volume.
Cancel resets to month 1
Pay per call with x402
Your agents can settle each isolated call without inventing a human account. The cost is published before execution and debited from your available credits. Perfect for agent-first flows without friction.
Affiliate coupons and codes
Your affiliate code is a unique code (not your account ID). Anyone registering with your code receives 500 bonus credits, and you also receive 500 bonus credits. Works between agents and LLMs too. Up to 3 verified users.
Built for humans and agents
Native MCP server
Typed tools for LLMs: list/create/move/update tasks, breakdown, sprint plan and AI generation, with Bearer atkr_ API-key auth.
Recursive Kanban
Tasks with subtasks up to 5 levels, dependencies, positions and drag & drop. Guarded against cycles and depth overflow.
AI planning
Gemini turns natural language into structured tasks, breaks work items down and optimizes sprints by team velocity.
Credit economy
Agent actions cost credits: buy packs, activate AutoRefill, or pay per call with x402. Manual task creation is always free.
Multi-language & themes
Dashboard in Spanish and English, light/dark/system theme. Local-first: your keys, your data.
Affiliate coupons
Invite 3 verified users with your code and both receive 500 bonus credits.
Credit pricing
Free
$0
1000 welcome credits · unlimited manual tasks
Credit packs
Small · 500 credits · $5
Credit packs
Medium · 1600 credits · $15
Credit packs
Large · 4500 credits · $40
Credit packs
Enterprise · 12000 credits · $100
One-time Stripe payments. No forced subscriptions.
Each credit covers one agent operation: task, active workspace, or active agent/apiKey per month.
Let your agent run the project
Set up the MCP server in 2 minutes and try conversational planning.