A shared task list for Cursor
By Jake Marsh, who builds Taskhold. Updated 2026-10-31. Facts about other tools checked 2026-10-31. 6 min read.
Cursor's agent is good at the task in front of it. What it doesn't have is a backlog that you control from outside the editor and that survives the end of a chat.
Taskhold gives it one. You add tasks in the web app; Cursor reads and updates the same list through MCP tools. The live board shows what it started, how far each task has got step by step, where it got stuck and what it finished, each change marked as yours or the agent's. Running Claude Code and Cursor on the same backlog works the same way, and the live board demo shows it without an account.
Add Taskhold in one click
- Add to CursorOpens Cursor on this computer, then asks you to connect. No key to copy and no sign-up first: your agent gives you a link to start the 7-day trial the first time it uses Taskhold.
Setup
A setup URL gets Cursor connected in one paste. Create one in Taskhold Settings, drop it into the agent chat with "set this up", and the agent adds the ~/.cursor/mcp.json entry with a new key, then confirms with taskhold_list. Links last 3 days; the quickstart has the long version.
Or connect with no key to copy: the Add to Cursor button at the top of this page opens Cursor with the entry filled in. You can also add just the URL to ~/.cursor/mcp.json yourself. Either way, open Customize in the sidebar (older versions: Cursor Settings > MCP), where Taskhold shows a button to connect. A browser window opens. If you're signed in to Taskhold you click Allow; if not, it connects right away, and the first time your agent uses Taskhold it gives you a link to start the 7-day trial or use your existing account.
{
"mcpServers": {
"taskhold": { "url": "https://taskhold.com/api/mcp" }
}
}To add it by hand with an API key, put this in ~/.cursor/mcp.json for every project, or .cursor/mcp.json in one project:
{
"mcpServers": {
"taskhold": {
"url": "https://taskhold.com/api/mcp",
"headers": { "Authorization": "Bearer <your atd_sk_ key>" }
}
}
}To keep the key out of the file, use "Authorization": "Bearer ${env:TASKHOLD_API_KEY}" and set that variable in your shell. Cursor may need a restart, or a toggle of the server in Customize, to pick it up.
Cursor asks for approval before it uses an MCP tool, so the first Taskhold calls each wait for a click. In the Auto-review run mode, tools on your allowlist run straight away; allowlisting the taskhold tools keeps progress reports from interrupting the agent.
Write the workflow as a Cursor rule
Cursor loads project rules from .cursor/rules, and only files ending in .mdc count there; a plain .md in that folder is ignored. Save this as .cursor/rules/taskhold.mdc. alwaysApply: true puts it in every chat instead of waiting for a matching file:
---
description: Taskhold task list workflow
alwaysApply: true
---
# Taskhold
Use the taskhold MCP tools for every request; the person follows along on a live board.
- Always send project: "<repo folder name>".
- Before anything else: taskhold_get_current and taskhold_updates. Hold on to the cursor.
- Starting work: taskhold_begin (title, project, steps, agent: "Cursor in <repo>"). For queued work, taskhold_list with status todo, then taskhold_begin with its task_id.
- After each step: taskhold_complete, start_next: true. Report long steps with taskhold_progress, no more often than every few minutes.
- Stuck on a decision: taskhold_block with one question, then poll taskhold_updates (task_id, since, wait: 50).
- Finished: taskhold_complete on the task, saying what changed and how you tested it.
- Bugs or TODOs left for later: taskhold_create, or taskhold_create_from_todo with file and line.Cursor also reads AGENTS.md in the project root and in subfolders, plain markdown with no frontmatter. If the repo already has one that Claude Code or Codex uses, paste the lines under the frontmatter there instead, and every agent follows the same rules.
See what Cursor's agent is doing without watching the editor
Cursor's agent panel shows the run you're looking at. The live board covers the ones you aren't: each task with its steps, the last progress message and anything blocked under Needs you. The monitoring guide explains the status dot and time estimates, and the MCP tools reference lists every call behind them.
The rule above passes a label such as "Cursor in web-app" when the agent starts a task. That label is what the live board shows next to the task.
Cursor plus another agent
Many people run Cursor next to Claude Code or Codex. All of them can connect to the same Taskhold list, each with its own key. A task one agent has started shows as in progress, so the others take the next one, and the live board shows each agent's tasks side by side.
Turning Taskhold off in Cursor
Cursor sends only what its agent writes into tasks: titles, steps, progress and notes. Switch the server off in Customize to pause it, or delete the taskhold entry from mcp.json to remove it. Revoking the key in Taskhold Settings > API keys stops that machine at once with 401.
A red dot or missing tools in Customize usually comes down to the JSON, the key or an environment variable Cursor can't see. Cursor MCP server not connecting goes through each case.
Answer your coding agents from anywhere.
$12/month or $120/year after the trial. Card up front. Cancel before day 7 and you pay nothing. Every agent you run is included.
Or connect your agent first, no sign-up: add https://taskhold.com/api/mcp as an MCP server. The first time it uses Taskhold, it gives you the trial link.
Questions
- Does it work in Cursor's background agents?
- Anything in Cursor that loads your MCP servers can use it. It is a plain remote MCP server with a bearer header, nothing Cursor-specific.
- Is my code sent anywhere?
- Only what the agent puts in a task: text, notes, and file paths with line numbers if it files a TODO. Taskhold doesn't read repositories.
- Can I use one key for Cursor and Claude Code?
- You can, but making one setup URL per client is cleaner: each gets its own key that you can revoke separately in Settings.
- Will Cursor report progress on its own?
- When the rule is in .cursor/rules or AGENTS.md, the agent calls taskhold_progress as part of the work. Without it the agent may only start and complete tasks, and the bar fills by steps done or not at all.
- What does it cost?
- One plan: $12 a month or $120 a year per person. The 7-day trial starts with a card.
Sources
- Cursor docs: MCP (mcp.json, headers, variables, approval), checked 2026-10-31
- Cursor docs: rules (.cursor/rules, .mdc, AGENTS.md), checked 2026-10-31
Related
- A shared todo list for Claude Code and Cursor
- How to see what Claude Code is doing, live
- Windsurf task list over MCP: watch progress live
- Answer a blocked coding agent from your phone
- Cursor MCP server not showing tools: the fix
- Cursor agent waits for input silently: the fix
- Cursor hook: show on the board when agent stops
Updated 2026-10-31. Something wrong or out of date? Email hello@greatwork.company.