What it is: Linear Agent is the AI agent built into Linear, launched in public beta on March 24, 2026. It answers questions about your issues, projects, documents, and code, creates and updates work, drafts specs and updates, and can open a pull request to fix an issue. Open it with Cmd/Ctrl + J, mention @Linear in a comment, or mention it in Slack or Microsoft Teams.
What it costs: the chat is included on every plan, including Free. Two features bill against prepaid AI credits on paid plans: Loops (background automations, typically $0.07 to $0.20 per run) and coding sessions (model tokens at cost plus $0.25 per 20 minutes of sandbox time).
Where it stops: it only sees what is in Linear. It can file issues from Slack, Intercom, Zendesk, and Gong, but not from HubSpot.
Linear Agent at a Glance
| Feature | What it does | Plan | Cost |
|---|---|---|---|
| Chat and @Linear | Ask questions, create and update work, draft documents | All plans, including Free | Included |
| Skills | Save a conversation as a reusable workflow | All plans | Included |
| Slack | Mention @Linear to turn a thread into an issue | All plans | Included |
| Intercom and Zendesk | Turn a support ticket into triage issues in one click | Business, Enterprise | Included |
| Triage Intelligence | Suggest team, assignee, labels, and duplicates on new issues | Business, Enterprise | Included |
| Loops | Run the agent on a schedule or on events | Paid plans | AI credits, about $0.07 to $0.20 per run |
| Coding sessions | Write and test code, return a draft pull request | Paid plans | AI credits: tokens at cost plus $0.25 per 20 minutes |
Three checks before you lean on it:
Check these first
- Who can spend AI credits? The balance is shared across the workspace, and Linear's docs say admins can't yet limit spending per user.
- Is the context in Linear? The agent works only from issues, projects, documents, comments, and code it can see. Customer conversations in another tool are invisible to it.
- Do you need Loops or coding sessions? If not, the agent costs nothing beyond your seats, and you can skip the credit setup entirely.
In this article
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
What Linear Agent Is
Linear Agent is Linear's native agent. Linear's docs describe it as able to "answer questions about your workspace and codebase, take actions for you, and connect with other tools in your stack." It works across teams, projects, cycles, milestones, initiatives, issues, comments, activity history, and documents, and it "operates within your existing permissions," so it can only read or change what the person asking can already see.
Linear announced it on March 24, 2026, as a public beta for all teams. It is on by default. At launch, Linear said the chat should stay in the base seat price after the beta and that high-volume features could become usage-based above a threshold. That is now the model: chat is included, while Loops and coding sessions bill against AI credits.
It is separate from third-party agents. Linear also lets you install outside agents from its Integrations Directory and assign them issues. Linear's docs say those agents are not billable seats, though the companies behind them may charge.
What Linear Agent Can Do
Answer questions about your work
Find information across issues, projects, documents, and code, and summarize threads and customer requests. Linear's launch post shows it pulling related issues into a new project and extracting common requirements from customer requests into a draft spec.
Create and update work
Create and update issues, projects, milestones, and initiatives from a chat or a comment, and draft documents and stakeholder updates.
Save skills
Save a conversation that worked as a personal or team skill. You run a skill with a slash command, and the agent can use it on its own when the context matches.
Run Loops in the background
Loops run the agent on a schedule or when something happens: an issue enters Triage, a project changes, a pull request gets a review, or a message lands in a chosen public Slack channel. Linear's own examples include investigating new bugs in Triage, summarizing a team's week, and drafting customer-facing messaging from the pull request that closed an issue.
Write code
Delegate a well-scoped issue and the agent starts a coding session. It writes and tests the change in a cloud environment using Claude Code or Codex, then returns a draft pull request. Your GitHub permissions and review rules still decide what ships.
Reach other tools through MCP
Admins can connect MCP servers, with Granola, Notion, and Figma named as examples, so the agent can pull outside context into chats, comments, and Loops.
Where You Can Use It
In the Linear app, open a chat with Cmd/Ctrl + J or at linear.app/agent, or mention @Linear in a comment on an issue, document, or update. The launch post says it also works in the mobile app.
Outside Linear, mention @Linear in Slack to turn a conversation into an issue, which Linear's October 2025 changelog lists on all plans. It is also available in Microsoft Teams. In December 2025 Linear added the agent to its Intercom and Zendesk integrations on Business and Enterprise: one click on a support ticket and it reads the whole conversation, files the bugs and feature requests it finds into the right team's triage queue, and attributes them to the customer. On Enterprise it does the same for Gong call transcripts with no clicks at all.
That list matters if your support team works somewhere else. More on that below.
Linear Agent Pricing
There is no separate Linear Agent subscription. You pay for Linear seats as usual, which is $10 per user per month on Basic and $16 on Business, both billed yearly (our Linear pricing guide has the full breakdown). Then two features draw from AI credits:
Loops
Available on paid plans. Linear's docs put a typical run without code at $0.07 to $0.20. A loop that starts a coding session also pays for that session.
Coding sessions
Available on paid plans. You pay model tokens "at provider-published rates, with no markup," plus sandbox runtime at $0.25 per 20-minute block. Clearly scoped issues cost less, because the agent spends less time exploring the codebase.
How AI credits work, from Linear's docs:
AI credits, the short version
- Credits are a prepaid workspace balance in US dollars. If you never add funds, you never get charged and those two features stay off.
- Admins add them in Settings > Workspace > Billing > AI Usage and credits. The minimum top-up is $10, and automatic reload starts at $50. Credit cards only.
- Purchased credits expire 12 months after purchase. When the balance runs out, Loops pause and new coding sessions can't start.
- Failed runs, retries, and partial work are billed for what they used. Added funds are non-refundable.
Linear's docs also say other AI features in your plan don't draw from credits, so Triage Intelligence and the chat itself don't add to the bill.
Linear Agent vs Triage Intelligence
They are easy to confuse. Triage Intelligence is a Business and Enterprise feature that looks at each new issue in Triage and suggests a team, project, assignee, and labels, plus likely duplicates and related issues. You can set each suggestion to show, hide, or auto-apply per team. It takes one to four minutes per issue.
Linear Agent is the assistant you talk to, and Loops are how you make it act without being asked. A triage Loop can go further than suggestions: investigate a new bug, write a customer impact summary, or delegate a fix to a coding session. Our Linear triage guide covers the queue itself.
How to Set Up Linear Agent
- 1
Decide whether to keep it on
It is enabled by default. To turn it off, a workspace admin or owner goes to Settings > AI & Agents > Linear Agent and switches off Enable Linear Agent.
- 2
Write workspace guidance
Add guidance in Settings > AI so the agent knows your conventions, such as how you write issue titles or which team owns what. Teams and individuals can add their own guidance on top.
- 3
Connect Slack and MCP servers
Update the Slack integration so people can mention @Linear in threads, and have an admin add any MCP servers you want the agent to read from.
- 4
Add AI credits, then a spend limit
Only if you want Loops or coding sessions. Add credits, decide whether to turn on automatic reload, and review spending under Settings > Usage & limits.
- 5
Grant code access for coding sessions
An owner or admin with GitHub organization owner access grants code access in the GitHub integration settings, enables coding sessions, and creates an environment for each repository.
- 6
Start with one Loop
A good first one is a triage Loop that investigates new bugs and adds a summary. Watch a week of runs and their cost before adding more.
Limits to Know Before You Rely On It
Where it is strong
- Included in every plan for chat and skills
- Works where people already talk: Linear, Slack, Teams
- Respects existing permissions
- Usage pricing for heavy features, with no markup on model tokens
Where to be careful
- Credits are shared, with no per-user spend limits yet
- Failed and partial runs are still billed
- Coding environment secrets are visible to anyone who can run a session there
- Issue-triggered Loops run only on issues created inside Linear by default
- Only sees context that lives in Linear or a connected source
Two more points from outside Linear's docs. The Register noted at launch that the announcements said little about security beyond permissions, and that agents acting on text written by others carry prompt injection risk. And Loops with web access can, in Linear's own words, "send workspace content to external services," so leave that setting off unless a loop needs it.
Getting HubSpot Tickets in Front of Linear Agent
Linear Agent is only as useful as the context in Linear. For teams on Intercom or Zendesk, Linear now files support conversations into triage with the customer attached. For teams whose support runs on HubSpot Service Hub, nothing arrives on its own: the bug report sits in a HubSpot ticket, and someone copies it into Linear by hand, if they remember.
IssueLinker closes that gap. From the HubSpot ticket, support creates a Linear issue in one click, optionally drafted by AI from the full ticket conversation with screenshots and files attached, or links an existing issue. On HubSpot Pro and above, a workflow can create the issue with nobody opening the ticket. IssueLinker can attach the ticket's company as a Linear customer request with revenue and tier from HubSpot, which is exactly the data Linear Agent summarizes when you ask which customers want something.
The loop back matters as much. Comments sync both ways, and the Linear status is written back to ticket properties, so when an issue moves to Done, whether a person or a coding session fixed it, support sees it in HubSpot. You can also post a note on the ticket or email the ticket owner when the issue closes. See the HubSpot Linear integration for the full feature list, or our guide to integrating Linear with HubSpot for the other options.
Get HubSpot tickets into Linear, and the status back
IssueLinker creates the Linear issue from the HubSpot ticket, attaches the customer, and syncs comments and status both ways. $39 per month flat, unlimited users, any HubSpot plan, 14-day free trial.
Who IssueLinker Is Not For
IssueLinker connects HubSpot tickets to Linear, Jira, and GitHub. If your support runs on Intercom or Zendesk, Linear's own integration with Linear Agent already covers ticket-to-issue on Business and Enterprise. If you want Linear Agent for planning and code only, with no support tickets involved, you don't need it either. See pricing for the current plans.
