Quickstart
Get from a half-formed idea to your first AI-executed task in about ten minutes. This guide assumes nothing — no existing project, no installed tools, no team.
Taskachu is an execution system, not another task tracker. That distinction matters for how you’ll use it. Where Linear stops at “task assigned, in progress, done”, Taskachu keeps going: it captures the idea, builds a knowledge base around it, hands it off to your AI coding tool, and waits for you to close the loop.
By the end of this page you’ll have a workspace, a structured set of documents, your first backlog of cards, and a handoff ready to send to Claude or Cursor.
Sign up & open the app
Taskachu is a web app — there’s no install. Head to app.taskachu.comand create a free account with email or Google. You’ll land on the Workspaces page with a starter workspace already prepared.
Free covers a single workspace with a monthly AI quota. Upgrade only when you want multiple workspaces or higher quotas — see pricing.
Create your first workspace
A workspace is one product. Most solo founders have one or two going at once — pick a name that you’ll still understand in six months. From the sidebar, click + New workspace, type a name, and you’re in.
Capture your idea
Inside your workspace, two main surfaces handle the capture phase:
- Documents — long-form Markdown pages with a parent-child tree. Drop your idea here as a brain-dump; the AI can later draft a problem statement, target audience, and rough architecture from it.
- Boards & Cards — once an idea is roughly shaped, run AI decompose on a card to fan it out into subtasks. The decomposed structure becomes your backlog.
Every document you create gets indexed for retrieval. When you later use Ask Taskachu or hand a card off to Claude/Cursor, the relevant chunks are attached automatically — you don’t re-paste context.
Pick your first task
Open any board and click Prioritize. The AI ranks your cards by impact and effort and surfaces the single top pick for today.
| Field | What it means | Set by |
|---|---|---|
priority | High / medium / low. | You or AI prioritise |
subtasks | Inline checklist. Generated by AI decompose. | AI · editable |
description | Free-form Markdown. The agent reads this verbatim. | You |
labels | Free-form workspace-scoped tags. | You |
Hand off via MCP
Taskachu speaks Model Context Protocol — your Claude Code, Claude Desktop, or Cursor agent can list, read, create, update, and move cards in your workspace exactly the way the UI can. Generate a workspace-scoped Personal Access Token in Settings → API Tokensand drop it into your AI tool’s MCP config.
{
"mcpServers": {
"taskachu": {
"type": "http",
"url": "https://app.taskachu.com/api/mcp",
"headers": {
"Authorization": "Bearer ${TASKACHU_PAT}"
}
}
}
}Restart your AI tool. The agent now sees your workspace, can read every card and document, and acts under the same identity and permissions as you do in the UI.
The token has the same permissions as your account inside the workspace it was issued for. Use environment variables, not committed config. Rotate from Settings → API Tokens any time.
Review and close the loop
Your AI tool returns a code change as a regular pull request. Review it like you would any PR. When you merge, mark the Taskachu card as Done — the activity log, comments, and completion state stay tied to the original card.
- Open the PR in your repo. Read the diff.
- If it needs another pass, leave a comment on the Taskachu card — your agent will see it on the next session.
- When you merge, drag the card to Done. The board updates and the next priority surfaces.