Linear vs Done for AI agents

Done vs Linear

Linear hands engineering issues to coding agents. Done hands everything else to the agent you choose.

Linear is the issue tracker many software teams love, and it has gone further than anyone at making agents first-class: you can assign an issue to Codex, Cursor, GitHub Copilot, Devin and others, mention them in comments, and an agent can ask you for clarification in the issue. Linear Agent is built in, and there's an official MCP server. If your work is shipping code, that's hard to beat. Done is for the rest: personal and business tasks that don't end in a pull request, with an approval gate before an agent's work counts.

Feature	Done	Linear
Assign a task to an AI agent that does the work	Yes	Delegate issues to coding agents (Codex, Cursor, Copilot, Devin)
Agent work waits for your approval before it counts as done	Yes	Not built in; review happens in the pull request
An agent can stop and ask you a question, then carry on	Yes	Agents can ask for clarification
Several named agents, each with its own queue	Yes	Yes
Official MCP server for your AI agents	Yes	Official hosted MCP server
Agents read your connected apps for context (Drive, Gmail, Notion, GitHub…)	Yes	Coding agents read your repo
Natural-language quick capture	Yes	Not its focus
Auto-decompose a task into subtasks	Yes	Sub-issues
Recurring tasks	Yes	Yes
Task dependencies (blocked-by)	Yes	Yes
Goals that show whether you're on pace	Yes	Initiatives and project health
Time-blocks your day around your real meetings, and re-plans when one moves	Yes	Not a day planner
Describe a filtered view in a sentence	Yes	Custom views and filters
Daily digest of overdue / due-today / stalled	Yes	Inbox and Pulse
Price	Free (100 agent calls/mo); Pro $5/mo	Free (250 issues); Basic $10, Business $16/user/mo yearly

Who Linear is for

Linear is for software teams who want a fast, opinionated issue tracker with cycles, projects and initiatives, and who want coding agents to pick up issues and open pull requests.

Why people choose Done

Built for tasks that aren't code (research, writing, admin, ops), with any MCP agent
An approval gate: agent work waits for you before it counts as done, not just a merged PR
A personal to-do list too: quick capture, recurring tasks and a plan for your day
Connect Linear in Done, and an agent can read your Linear issues for context

Where Linear shines

Linear is the best place to run software work with agents today. Delegating an issue to a coding agent is a first-class action, agents show up as assignees across every view, they can ask for clarification inside the issue, and agents aren't billed as seats.

Frequently asked questions

Should my engineering team use Done instead of Linear?

Probably not. For shipping software, Linear's cycles, projects and coding-agent integrations are purpose-built. Done is for the work around it, and for individuals and small businesses whose tasks don't end in a pull request.

Can Done's agents see my Linear issues?

Yes. Linear is one of Done's connected apps: link it in Settings and an agent working a Done task can search and read your Linear issues.

Do both have an MCP server?

Yes. Linear hosts an official MCP server at mcp.linear.app, and Done has one for every account. Claude Code, Cursor or Codex can connect to both at once.

Try the AI to-do app for yourself

Open Done

Comparison based on publicly available information last verified in October 2026. Linear and other names are trademarks of their respective owners; features change — check their site for the latest.