Intro to Linear Agents + Claude

Use Linear Agent and Claude Desktop to manage software projects with AI.

Introduction

30 Second Summary

Imagine you just joined a startup as a project manager. The team shipped a mobile app six months ago and the backlog is chaos: bug reports from users, feature requests from the CEO, performance complaints from QA. What if AI could sort through all of it so you could focus on making decisions instead of organizing tickets?

In this project, you will set up an AI-powered project management workflow using Linear. You will create a realistic project board, then use AI assistants to manage your work through natural conversation.

What You'll Build

A complete AI-powered project management system using Linear, Linear Agent, and Claude Desktop connected via MCP for natural language project management.

By the end of this project, you'll have:

🏡 A Linear workspace with a team, project, and sample issues organized for a simulated software project.
🤖 Linear Agent configured to create, update, and summarize issues through natural conversation.
🚦Triage Intelligence enabled (via a Business plan trial) to detect duplicate issues and suggest smart routing.
🧠 Claude Desktop connected to Linear via MCP so you can manage your project board from outside Linear.
💎 Secret Mission: Create a custom Linear Agent Skill that automates recurring workflows like weekly status reports.

Are there any prerequisites?

You need a computer running macOS or Windows, internet access, and a free Linear account. Claude Desktop is not available on Linux. Most of this project uses Linear's free plan.

Step 5 (Triage Intelligence) requires a Linear Business plan. Linear offers a free trial of the Business plan, which is enough to complete this step at no cost.

Set Up Your Tools

Before you can manage a project with AI, you need two tools ready to go.

Linear is where your project lives. Claude Desktop is the AI assistant you will connect to Linear later so you can manage work through natural conversation.

In this step, get ready to:

Set up a Linear workspace for project management.
Install Claude Desktop for AI-powered conversations.
Verify both tools are ready to use.

Set up your Linear account

Linear is the project management tool where you will track issues, organize work, and interact with AI features like Linear Agent.

To get started with Linear, you need a workspace. Pick a name for yours (e.g. your name, a company name, or a team name).

Enter your workspace name here: your workspace name

✔️ I already have a Linear account

Confirm you can see your workspace dashboard.

...and you already have a team in sidebar showing your teams.

If you can see at least one team listed in the left sidebar, you are good to go.

ⓧ I need to create a Linear account

Go to linear.app and click Sign up.

When prompted for a workspace name, enter your workspace name.

Do you see "This workspace URL is already taken"?

No worries! Just add your name to the end of your workspace name - or any other unique twist to make it a one-of-a-kind.

Select Create Workspace.
Welcome to Linear!

Select Get started.
Choose your theme of choice.

Select Continue

Select Continue to pass through the keyboard shortcuts.
Select I'll do this later to bypass GitHub integrations.

When it asks to invite co-workers to your team, select I'll do this later.

Leave the Subscribe to updates options as they are - we can always change them later.
Select Continue

Select Open Linear

You made it! Welcome to Linear's default starting screen, in your new workspace.

Wait - what actually is a workspace?

A workspace is the top-level container in Linear. It represents your entire organization and holds all your teams, projects, and issues. Think of it as your company's home base in Linear.

Linear will also create a default team for you automatically. Can you spot it in the left menu?

Cool! What's a team?

A team is a group within your workspace that owns its own issues, workflows, and labels. A workspace can have multiple teams (e.g. Engineering, Design, Marketing). When you created your workspace, Linear automatically created a default team with the same name.

Epic - made it into Linear. Now we can get started with Claude Desktop, the second tool we'll be needing for this project.

Before we move on...is Linear really free?

Linear's free plan covers Linear Agent and MCP support, which is what most of this project uses. One step (Triage Intelligence) needs a Business plan. Linear offers a free 14-day trial with no credit card, which is enough to finish the project at zero cost.

The free plan supports up to 250 issues and 2 teams. That is more than enough for this project.

Set up Claude Desktop

Claude Desktop is the app you will connect to Linear later in the project. It lets you manage your project board through natural language conversations.

✔️ I already have Claude Desktop

macOS

Press Cmd+Space to open Spotlight.
Type Claude and press Enter to open it.

Navigate to the chats screen

Type Hello in the message box at the bottom of the window and press Enter.

Claude should reply within a few seconds. If you see a response appear below your message in the chat window, Claude Desktop is ready.

Windows

Press the Windows key to open the Start menu.
Type Claude and press Enter to open it.
Type Hello in the message box at the bottom of the window and press Enter.

Linux

Claude Desktop is not available on Linux

Claude Desktop currently supports macOS and Windows only. You can still complete the Linear-only steps of this project in your browser.

ⓧ I need to install Claude Desktop

macOS

Go to claude.ai/download and click macOS.
Open the downloaded .dmg file and drag Claude to your Applications folder.
Press Cmd+Space to open Spotlight, type Claude, and press Enter to launch the app.
Sign in with your Anthropic account. A free account works perfectly.

macOS blocking the app from opening?

Go to System Settings, then Privacy & Security, and click Open Anyway next to the Claude message.

help me install Claude Desktop on macOS

Windows

Go to claude.ai/download and click Windows.
Open the downloaded installer and follow the prompts to complete installation.
Press the Windows key to open the Start menu, type Claude, and press Enter to launch the app.
Sign in with your Anthropic account. A free account works perfectly.

Windows blocking the installer?

If Windows Defender SmartScreen shows a warning, click More info, then click Run anyway to proceed with the installation.

help me install Claude Desktop on Windows

Linux

Claude Desktop is not available on Linux

Claude Desktop currently supports macOS and Windows only. If you are on Linux, you will not be able to complete the MCP connector steps (Steps 5 and 6) of this project.

You can still complete Steps 1 through 4 using Linear's built-in AI features, which work entirely in the browser.

Do I need a paid Claude account?

No. The free Claude plan supports the Connectors feature you will use to connect Linear later. You do not need Claude Pro for this project.

Your tools are ready. Next up, you will create a project and fill it with realistic issues so the AI features have something meaningful to work with.

Create Your Project and Issues

With Linear and Claude Desktop in place, it is time to build something real. Every project starts with a backlog of work: bugs to fix, features to build, and improvements to ship.

Your first task: get everything into Linear so the team has a single source of truth. You will create a project, populate it with realistic issues, and organize them with labels. Once the backlog exists, you can let the AI features do the heavy lifting.

In this step, get ready to:

Create a project in Linear to organize your work.
Use Linear Agent to batch-create issues and apply labels.
Verify your project board has a realistic backlog.

Create a new project

Your workspace is the home base. Now you need a project inside it to organize a specific piece of work. A workspace can hold many projects over time, but each project focuses on one goal.

✔️ I have my own project idea

Awesome!

Enter your project name here: PROJECT_NAME.
You will use this as your Linear project name throughout this step.

ⓧ I don't have a project idea yet

No worries. We will use a ready-made scenario: picture this!

You are the new project manager at a startup that just shipped a mobile fitness app. So healthy. Bug reports, feature ideas, and performance complaints are scattered across Slack messages and sticky notes.

Copy this project name by clicking on it: Mobile App Redesign.
Paste the copied variable name here: PROJECT_NAME
Now it will show up for you automatically in later instructions in the project!

Onwards...

Remind me, what is a project in Linear? What about "issues"?

A project groups related issues under a single goal with a timeline. Think of it as a container for all the work needed to ship one feature, fix one problem area, or complete one initiative.

Issues live inside projects and represent individual tasks. A project can span multiple weeks and contain dozens of issues, while each issue is a single piece of work someone can pick up and complete.

In the Linear sidebar, click Projects to open your workspace project list.

Click the + button at the top of the project list to open the project creation dialog.

Enter your project or app name as the project name.

Add a one-sentence description of what your project aims to accomplish. For example Redesign the mobile app for better performance and user experience

Set the project status to In Progress.

Click Create project to save it.

Look for your new project - can you see in the sidebar under your team?

Don't see the project creation dialog?

You can also press Cmd+K (macOS) or Ctrl+K (Windows) to open the command menu, then type Create project to find the option. Make sure you are inside a team workspace (not the home view).

help me find where to create a project in Linear

Still stuck? Ask for help in the NextWork community.

Create your project's issues

Before you let the AI take over, create one issue manually so you understand how Linear's issue system works.

Think of a real task from your project, or use the example below. Got it? Off we go!
Press C anywhere in Linear to open the issue creation form.

Enter a title for your first issue. If you are using the example project, type App crashes on login when using biometric authentication.

Set the priority to Urgent.

In the project field, select your project (your project or app name).

Click Create issue to save it.

Click your project name in Projects view.

At the top of your Project page, look for the Issues bubble.

Click Issues.

You should see the issue you just created listed with its Urgent priority badge.

A real project needs more than one issue. Creating them one-by-one would be tedious. This is exactly the kind of repetitive work AI is built for.

Press Cmd+J (macOS) or Ctrl+J (Windows) to open the Linear Agent chat.

Having trouble with keyboard shortcuts?

You can also select the Ask Linear chatbot icon in the bottom right corner of the screen.

Still stuck? Ask for help in the NextWork community.

Choose the prompt that matches your project:

✔️ I'm using my own project

Copy and paste this prompt into the Linear Agent chat:

Create 5 issues in the [[PROJECT_NAME="your project or app name"]] project with a mix of bugs, features, and improvements. Give each issue a different priority level (Urgent, High, Medium, or Low).

Press Enter to send the prompt.
Wait for Linear Agent to confirm it created all five issues.

ⓧ I'm using the example project

Copy and paste this prompt into the Linear Agent chat:

Create 5 issues in the Mobile App Redesign project with these titles and priorities:
1. Redesign onboarding flow for new users - High priority
2. Add dark mode support - Medium priority
3. Loading spinner stuck on home screen after update - High priority
4. Improve push notification delivery speed - Low priority
5. App takes too long to open on older devices - Medium priority

Press Enter to send the prompt.
Wait for Linear Agent to confirm it created all five issues.

What does this prompt do?

You are asking Linear Agent to batch-create five issues in a single message, each with a specific title and priority level. Instead of opening the creation form five separate times, the agent parses your list and creates each issue in your project automatically.

Expect the agent to confirm each issue it created, listing the title and priority. If it asks for clarification (like which team), answer and it will proceed.

Navigate to your your project or app name issues page.
Confirm all six issues (the one you created manually plus the five from the agent) are listed.

Check the pirorities on each issue match your prompt.

What just happened?

You created five issues in about ten seconds. Doing this manually would have meant opening the creation form five times, typing each title, selecting each priority, and choosing the project for each one. This is the core value of conversational AI for project management: bulk operations through natural language.

🙋‍♀️ Linear Agent didn't create all 5 issues?

If the agent only created some of the issues, try sending a follow-up message asking it to create the missing ones. If it refuses or misinterprets the request, create the remaining issues manually using the same process you used for the first one.

help me batch-create issues with Linear Agent

Add labels and context

Labels help both humans and AI categorize work at a glance. When the AI sees patterns across labeled issues, it can make smarter suggestions about routing and prioritization.

Hover over your team name in the sidebar.
Click the three dots that appear, then select Team settings.

In the settings menu on the left, click Issue labels.

Click the + New label button at the top of the label list.

Type Mobile Only Bug as the label name and click Enter.

Repeat to create three more labels: Tech Debt, Performance, and UX.

Now use Linear Agent to apply labels to all six issues at once instead of opening each one individually.

Navigate back to the app by selecting Back to app in the top left.

Press Cmd+J (macOS) or Ctrl+J (Windows) to open Linear Agent.
Choose the prompt that matches your project:

✔️ I'm using my own project

Copy and paste this prompt into the Linear Agent chat:

Look at all the issues in my [[PROJECT_NAME="your project or app name"]] project and apply one of these labels to each: Bug, Feature, Performance, or UX. Pick the label that best fits each issue.

Press Enter to send the prompt.
Wait for Linear Agent to confirm it applied a label to each issue.

ⓧ I'm using the example project

Copy and paste this prompt into the Linear Agent chat:

Look at the labels available and label issues in Mobile App Redesign appropriately

Press Enter to send the prompt.
Wait for Linear Agent to confirm it applied a label to each issue.

What does this prompt do?

You are asking Linear Agent to apply labels to each issue based on it's own judgment. The agent matches each issue title to a label and updates them all in one pass.

Expect the agent to confirm each label assignment. After it finishes, every issue in your project will be categorized, which helps both humans and AI filter and prioritize work.

Click on any issue in your project view to confirm the label appears in the issue detail sidebar.

Labels not appearing on your issues?

If Linear Agent could not apply the labels, you can do it manually. Open any issue, find the Label field in the right sidebar panel, and click it to select from your available labels.

You can also right-click an issue in the list view to find the label option in the context menu.

help me find where to add labels to issues in Linear

Still stuck? Ask for help in the NextWork community.

Your project is set up with a realistic mix of categorized issues. Next up, you will put Linear Agent to work and see how AI can manage this backlog through natural conversation.

Use Linear Agent to Manage Your Work

You have a realistic project backlog with issues, priorities, and labels. Now comes the real power of Linear: managing all of it through natural conversation instead of clicking through menus.

Linear Agent is a built-in AI assistant that understands your workspace context. You can ask it questions, create issues, add comments, and get summaries without ever leaving the chat window.

In this step, get ready to:

Open Linear Agent and query your project issues.
Ask for a strategic project summary and risk analysis.
Get AI-generated prioritization advice for your backlog.

Open Linear Agent and ask a question

You have already used Linear Agent to create issues and apply labels. Now you will use it to ask questions about your project and get strategic advice back.

Press Cmd+J (macOS) or Ctrl+J (Windows) to open the Linear Agent chat.
Copy and paste the following prompt into the chat:

What issues are currently in my [[PROJECT_NAME="your project or app name"]] project?

What does this prompt do?

You are asking Linear Agent to look up all issues inside your project and list them. The agent queries your workspace data and returns each issue with its title, priority, and current status.

Expect a structured list of your six issues (the one you created manually plus the five from the batch prompt). This confirms the agent has full visibility into your project.

Press Enter to send the message.
Check out the result - do you see all your issues?

What happened here?

The Linear Agent processes your question and returns a list of your issues with their priorities and statuses. It pulls this directly from your workspace data.

Still stuck? Ask for help in the NextWork community.

Get AI-generated summaries

Beyond creating and updating issues, Linear Agent can analyze your entire project and give you strategic summaries. This is especially useful for status updates and sprint planning.

Find the same open Linear Agent chat you were just in.
Copy and paste the following prompt:

Summarize the current state of the [[PROJECT_NAME="your project or app name"]] project. What are the biggest risks?

What does this prompt do?

You are asking Linear Agent to analyze all issues in your project and produce a strategic summary, including the biggest risks. The agent considers priorities, issue types, and overall backlog health to generate its assessment.

Expect a structured response covering the project status, a breakdown by category or priority, and a risk analysis highlighting urgent or blocking items.

Press Enter.
Once it's finished running, select the full screen icon in the top right of the Agent chat so we can read the answers more easily.

Read through the answer from the Linear Agent - what do you not agree with?

Now try asking for prioritization advice by copy and pasting the following to the agent:

Which issues should I prioritize this week?

What does this prompt do?

You are asking Linear Agent for prioritization advice. The agent evaluates your backlog by weighing urgency, issue type (bugs vs. features), and dependencies to recommend what to tackle first this week.

Expect a ranked list of recommendations with reasoning. The agent uses the same signals a project manager would: urgent bugs before low-priority features, blockers before nice-to-haves.

Press Enter.
Once it has finished responding, read through the response.
Can you see the issue marked with the Urgent priority?

Notice how the agent considers urgency, dependencies, and issue categories when making its recommendations. It is working with the same context you would use to make these decisions manually.

Get better results with specific prompts

The more specific your prompt, the more useful the response. Instead of "What's happening?" try "What urgent bugs are blocking the release?" or "Which features have no assignee?" Specific questions get actionable answers.

You have just experienced how conversational AI transforms project management. Next, you will see how Linear's Triage Intelligence works behind the scenes to automatically detect duplicate issues and suggest smart routing.

See Product Intelligence in Action

See Triage Intelligence in Action

You have been using Linear Agent to create and manage issues through conversation. That is powerful for individual tasks, but what about the patterns hiding across your entire backlog?

Linear's Triage Intelligence watches your workspace in the background and surfaces insights you would miss manually. It detects duplicate issues, suggests related work, and learns from your decisions over time. In this step, you will trigger it yourself and see how it responds.

In this step, get ready to:

Verify Triage Intelligence is enabled in your workspace settings.
Create duplicate issues to trigger AI-powered detection.
Review and act on Triage Intelligence suggestions.

Enable Triage Intelligence

Triage Intelligence requires a Business plan (Linear offers a free 14-day trial). Once your plan is active, you can enable the feature in your workspace AI settings.

Click your workspace name in the top-left corner of the Linear sidebar.
Select Settings from the menu that appears.

In the left-hand settings menu, scroll down and click Billing.
Look for which plan your workspace is on.

✔️ I see Business or Business Trial

You are all set. Skip ahead to the next instructions below to enable the AI settings.

ⓧ I see a different plan

Click View all plans next to your current plan.

Select Start a free trial under the Business plan.

Select Start trial.

Confirm the trial is active by navigating back to Billing.

Refresh your page.
Look for the confirmation you're on a trial of the business plan.

Will I be charged?

No. Linear's Business trial lasts 30 days and does not require a credit card. You can cancel or downgrade anytime before the trial ends.

Awesome! Upgrade success - let's check out Triage Intelligence.

Staying in your settings, scroll down to AI & Agents

Click on the Triage Intelligence setting.

Make sure the toggle for Enable Triage Intelligence is on.

Did you notice that the Behaviour rules all changed to Show?

What do these Behaviour rules mean?

Each Behaviour rule corresponds to a property that Triage Intelligence can suggest for incoming issues: team, project, assignee, and labels. Show means the AI surfaces its suggestion for you to review and accept or dismiss, without automatically changing anything.

The other options are Auto-apply (automatically accept suggestions without asking) and Hide (suppress suggestions entirely). Starting with Show keeps you in control while you learn what the AI recommends.

Create duplicate issues to trigger detection

Your backlog already has several issues from the previous step. You are going to create two new issues that describe similar problems in different words. This gives Triage Intelligence something to detect.

From your settings page, navigate back to the app by selecting Back to app in the top left.

Press Cmd+J (macOS) or Ctrl+J (Windows) to open Linear Agent.
Ask the agent to create your first duplicate issue by pasting the following prompt:

Create a new issue in my [[PROJECT_NAME="your project or app name"]] project titled "Login crashes when using Face ID"

Why are we creating duplicate issues?

This new issue describes the same problem as an existing one in your backlog, just with slightly different wording. Triage Intelligence works by comparing incoming issues against your existing backlog and flagging semantic overlaps.

By creating these intentional duplicates, you give the AI something concrete to detect. In a real workspace, duplicates sneak in when multiple team members report the same bug independently.

Press Enter.
Wait for the agent to confirm it has created the issue.

Now ask the agent to create a second duplicate issue.

In the same Linear Agent chat, paste the following prompt:

Create a new issue in my [[PROJECT_NAME="your project or app name"]] project titled "Face ID broken - not sure why"

Press Enter.

Great! We have two duplicate issues now. Let's see Triage Intelligence in action.

Open your Face ID broken - not sure why issue.

With the issue open, press Cmd+K (macOS) or Ctrl+K (Windows) to open the command palette.

Type Find Suggestions and select it from the results.

Look for the Triage Intelligence banner on your issue.

Wait for it to finish loading - see what it suggets!

Hover your mouse over the Related to issues.
Select Apply on the most similar issue - you choose!

In the Activity section of your issue, near the bottom, you should see your activity recorded.

On the right hand pannel of your issue, you might also see a new Relations section.

This is where any potential duplicates identified by Triage Intelligence will show up.

So cool! Triage Intelligence now enriches this issue in the background, comparing it against your existing backlog to find duplicates and related work.

You can use "Find Suggestions" on any issue

Triage Intelligence can run on issues in any status, not just ones sitting in Triage. Whenever you want to enrich an issue with duplicate detection and related-work suggestions, open it and search for Find Suggestions via the command palette.

🙋‍♀️ Not seeing any duplicate suggestions?

Triage Intelligence is a technology preview and detection may not be instant. Try waiting a minute and refreshing your issue list. You can also open the new issue detail view to check if suggestions appear there.

If nothing appears after a few minutes, the feature may still be indexing your workspace. Continue to the next step and check back later.

help me get Triage Intelligence duplicate detection working

Your workspace is now actively using AI to keep your backlog clean and connected. Next up, you will connect Linear to Claude Desktop so you can manage your entire project board through natural language without even opening Linear.

Connect Linear to Claude Desktop via MCP

So far, you have been managing your project from inside Linear using Linear Agent. That is powerful, but what if you could manage your Linear workspace without even opening it?

In this step, you will connect Claude Desktop to your Linear workspace using MCP. This lets Claude read and write Linear data directly, so you can create issues, check project status, and get summaries from a single conversation window.

In this step, get ready to:

Open Claude Desktop's connector settings.
Add and authorize the Linear connector.
Test the MCP connection by querying Linear from Claude.

Open Claude Desktop connector settings

Think of Connectors as Claude's address book for other apps. You add a service to the list, give Claude permission to talk to it, and from that point on Claude can read and update that service for you.

✔️ I still have Claude open

Fantastic. On we go!

ⓧ I need to open Claude again

Press Cmd+Space (macOS) or the Windows key (Windows) to open your search bar.
Type Claude and press Enter to open the app.

Click Claude in the top-left menu bar of the app.
Click Settings.

In the settings sidebar, click Connectors.

Don't worry if your connectors page is empty or has different apps to the screenshots. Everyone has different setups!

What's an MCP? Is it the same as Connectors?

MCP (Model Context Protocol) is the behind-the-scenes technology that makes Connectors work. Think of it as a shared language that lets Claude talk directly to apps like Linear.

Once connected, Claude can do anything you can do in Linear — search issues, create new ones, add comments. It uses your account permissions, so it never sees more than you would.

Add the Linear connector

Let's find Linear! And connect it to Claude.

Click the Browse connectors button to open the connector directory.

In the search bar, type Linear.

Click on the Linear connector to view its details.

Click Connect to start the authorization flow.

A browser window will open asking you to authorize Claude to access your Linear workspace.
Click Approve in the browser to grant access.

Select your workspace to connect.

You might see a pop-up asking if you'd like to return to Claude.
Select Open Claude.

It didn't take me back to Claude!

No worries - you can do it the old fashioned way and just find the Claude app again on your computer, and open it.

Back on your Connectors page, look for Linear.

Congratulations! You've successfully connected Linear and Claude.

What happened during authorization?

Behind the scenes, Claude used OAuth 2.1 to securely connect to Linear's MCP server at mcp.linear.app/mcp. This means Claude never stores your Linear password. It receives a limited access token that lets it interact with your workspace on your behalf.

🙋‍♀️ Don't see Linear in the directory?

Make sure you are using the Claude Desktop app, not the web browser version of Claude. The Linear connector is available on Claude Free and Pro plans in the desktop app.

If you still cannot find it, try clicking the "+" button from a chat instead. Click the "+" in the lower left corner of your chat, hover over Connectors, then select Manage connectors.

help me find and connect Linear in Claude Desktop

Test the connection

The best way to confirm everything is working is to ask Claude a question that requires Linear data.

Close the Settings window by clicking the back arrow in the top-left corner.

Click the New Chat icon in the sidebar, or press Cmd+N (macOS) / Ctrl+N (Windows) to start a fresh conversation.

Type the following message and press Enter:

What's going on in Linear right now?

What does this prompt do?

You are testing the MCP connection by asking Claude to pull live data from your Linear workspace. Claude uses the Linear connector to query your issues and return them in a conversational format.

Expect Claude to list the issues from your Mobile App Redesign project with their titles and priorities. If this works, your MCP connection is healthy and ready for more complex operations.

You may be prompted to give Claude permission to list issues from Linear.
Select Always allow.

Claude should return a list that includes the issues from your Mobile App Redesign project.

Great success! Now we can have some fun.

Connection failing or no issues returned?

Remote MCP connections are still early and the connection may fail or require multiple attempts. Try these fixes:

Go back to Customize > Connectors in Settings and disconnect Linear, then reconnect it.
Quit Claude Desktop completely and relaunch the app.
Make sure you authorized the correct Linear workspace (the one with your Mobile App Redesign project).

my Claude Desktop MCP connection to Linear is not working

Still stuck? Ask for help in the NextWork community.

Your Linear workspace is now connected to Claude Desktop. Next up, you will put this connection to work by managing your entire project board from Claude.

Manage Linear from Claude

In the previous step, you connected Claude Desktop to your Linear workspace via MCP. Now you have a powerful external AI assistant that can read and write to your project board without ever opening Linear.

In this step, you will put that connection to real use. You will query your workspace, create new issues, and add comments entirely from Claude Desktop.

In this step, get ready to:

Query your Linear workspace from Claude Desktop using multiple prompts.
Create a new issue and add a comment entirely from Claude Desktop.

Query your workspace from Claude

The real power of the MCP connection is that Claude can pull live data from your Linear workspace and present it conversationally. You do not need to navigate menus or remember where things are.

In Claude Desktop, click the New Chat icon in the sidebar, or press Cmd+N (macOS) / Ctrl+N (Windows) to start a fresh conversation.
Type the following prompt and press Enter:

What is the status of my [[PROJECT_NAME="your project or app name"]] project in Linear?

What does this prompt do?

This asks Claude to look up your specific project and summarize its current state. Claude uses the MCP connection to query Linear's API, then presents the project status, issue counts, and progress in a readable format.

Awesome! Claude has responsed with a summary of the project, including its status and current issues.

Try this next prompt:

Show me all urgent bugs in Linear

What does this prompt do?

This filters your workspace by both priority (urgent) and issue type (bug). Claude translates your natural language into the appropriate API query, fetches matching issues, and presents them as a focused list.

Claude should return your urgent priority issues from the workspace.

What is happening behind the scenes?

Each time you ask Claude about your Linear workspace, it uses the MCP connection to query Linear's API on your behalf. Claude translates your natural language into the right API calls, fetches the data, and then formats the response conversationally.

🙋‍♀️ Claude not returning your issues?

Make sure you are using the Claude Desktop app, not claude.ai in your browser. The web version requires a Team or Enterprise plan for connectors.

If Claude says it cannot access Linear, go to Settings > Connectors in Claude Desktop and check that Linear is still listed as connected. Try disconnecting and reconnecting if needed.

help me troubleshoot my Claude Desktop connection to Linear

Create and update issues from Claude

Reading data is useful, but the MCP connection also lets you write to Linear. You can create issues, add comments, and update properties without switching apps.

In Claude Desktop, copy and paste this prompt to create a new issue:

Create a new issue in Linear: Add haptic feedback to button interactions. Set priority to medium and add it to the [[PROJECT_NAME="your project or app name"]] project.

You may need to approve Claude's access to Linear again - if so, select Always allow.

What does this prompt do?

This instructs Claude to create a new issue in your Linear workspace with a specific title, priority level, and project assignment. Claude uses Linear's MCP tools to write data, not just read it.

Claude should confirm it created the issue and show you the details.

Confirm it exists by clicking on the hyperlink to the issue in Claude's response.

Woohoo! How easy was that?!
Back in Claude Desktop, type this prompt to add a comment:

Add a comment to the haptic feedback issue saying: Let's start with the primary action buttons first.

What does this prompt do?

This asks Claude to find the issue you just created by its title and append a comment. Claude resolves the issue reference, then uses the MCP write tools to add the comment on your behalf.

Claude should confirm the comment was added.

Jump back into Linear, refresh the haptic feedback issue and confirm the comment "Let's start with the primary action buttons first." appears. Like magic.

Why manage Linear from outside Linear?

Because Claude can talk to Linear on your behalf, you can turn anything into a ticket without leaving your conversation. Paste in a Slack thread and ask Claude to file it as a bug. Drop in meeting notes and have Claude extract every action item into separate issues.

Copy a customer support email and let Claude create a prioritized ticket with all the context included.

🙋‍♀️ Issue not appearing in Linear?

Check Claude's response carefully. If it mentions a permission error, your authorization may have expired. Go to Settings > Connectors in Claude Desktop and reconnect Linear.

If Claude says it created the issue but you cannot find it, make sure you are looking in the correct project. Click on your project name in your Linear sidebar to filter the view.

help me figure out why my issue created in Claude is not showing up in Linear

You can now read and write to your project board without ever opening Linear. Next up is the Secret Mission, where you will create a reusable Skill in Linear Agent that automates a recurring workflow like weekly status reports.

Secret mission

Create a Custom Linear Agent Skill

You have been using Linear Agent for one-off questions and actions. But what about tasks you repeat every week? In this secret mission, you will create a reusable Skill that automates a recurring workflow like generating a weekly status report.

Clean Up Your Resources

Decide whether to keep your resources running, pause them to come back later, or delete them entirely.

This project runs entirely on free tools, so there are no ongoing costs regardless of what you choose.

Resources you used:

Linear workspace (free plan) with a team, project, and sample issues.
Linear Agent chat and any saved Skills.
Claude Desktop app with the Linear MCP connector.
Triage Intelligence settings and duplicate detection data.

Keep everything running

No action needed. Choose this if you are still actively exploring Linear and Claude Desktop together.

Your Linear workspace, project, and issues will remain exactly as you left them.
The Claude Desktop MCP connection to Linear stays active.
Any Skills you created in Linear Agent remain available.
Linear's free plan has no time limit, so nothing will expire.

Pause - I'll come back to this later

Disconnect the MCP integration to keep your setup tidy while you are away.

In Claude Desktop, go to Settings, then Connectors.
Disconnect the Linear connector by clicking the disconnect option next to it.
Your Linear workspace and issues stay intact. You can reconnect anytime by re-authorizing the connector.

Delete - I don't want to use this again

Remove all project resources and start fresh.

In Claude Desktop, go to Settings, then Connectors, and disconnect the Linear connector.
In Linear, open your project "Mobile App Redesign" and delete it from the project settings.
Delete the sample issues you created by selecting them in the issue list and using the bulk delete option.
If you want to remove the workspace entirely, go to your workspace settings and select the option to delete the workspace.
If you no longer need Claude Desktop, drag it from your Applications folder to the Trash.

Nice Work!

You just built a complete AI-powered project management workflow from scratch. You now have two AI assistants ready to help you manage software development work through natural conversation.

You've learned how to:

Set up a Linear workspace with a realistic software project, team, and issue backlog.
Used Linear Agent and Triage Intelligence to create, summarize, and manage issues through natural conversation while detecting duplicates automatically.
Connected Claude Desktop to Linear via MCP to manage your project board from outside Linear using natural language.
Created a custom Linear Agent Skill to automate recurring status reports for your project.

Ready to quiz yourself?