[Day 8 of 30] Let Your AI See and Update Your Backlog
Add Atlassian or Linear MCP so your AI sees epics, stories, and tasks. One context, full product workflow.
The situation
You have calendar, email, and a capture habit. Your product work still lives in Jira, Linear, or Confluence. You copy titles into chat, paste links, describe context by hand. The AI cannot see your backlog, create issues, or update status.
Today we fix that. Add an MCP for your product tools and the AI gets structured access: epics, stories, tasks, pages. You ask in natural language; it creates, lists, and updates. No more switching tabs to copy ticket IDs.
This post continues the 1% Better series. If you have not read the previous posts, you can find them here.
The 1% upgrade
Connect Atlassian (Jira + Confluence) or Linear, the two most popular product tool stacks. Pick the one you use. One MCP, one auth, and your AI can work with your backlog.
Path A: Atlassian (Jira, Confluence, Compass)
MCP from Atlassian - OAuth in the browser, no API keys. Best if you use Jira Cloud, Confluence, or both.
Path B: Linear
Local MCP, one API key from Linear. Best if you use Linear for issues and projects.
After you connect, you’ll be able to create a basic tickets structure: Epic -> Story -> Task (or the equivalent in your tool). Then expand the list of your use cases with the examples below.
How to add the Atlassian MCP (Path A)
In Cursor (recommended flow):
Open the Cursor MCP Directory.
In the search box, type Atlassian and find the Atlassian server (Jira, Confluence).
Click Add to Cursor. The install screen for the MCP server in Cursor opens.
Click Install. Cursor adds the server to your MCP config.
Authorize: In a new chat, ask: “List my recent Jira issues” or “Show my Confluence spaces.” A browser window opens for Atlassian sign-in. Sign in and allow access. The token is saved; you do not need to do this again. Or open Settings -> Cursor Settings -> in the menu choose Tools and MCP, find Atlassian in the list and click Connect. Sign in in the browser when it opens.
Use it. Ask for issues, create stories, search Confluence, or create pages. The AI uses the Atlassian tools.
If Cursor asks to restart after install, restart once. Then try step 5 in a fresh chat.
In practice, authorization can sometimes drop after several weeks of use; Cursor may then fail to find your project or behave oddly. If you see that, open Settings -> Cursor Settings -> Tools and MCP, find Atlassian and click Connect again. Sign in in the browser when it opens. That usually fixes it.
In Claude Code (Connectors): Open Settings -> Connectors -> Browse Connectors, find Atlassian and click the + (plus) to add it. Complete OAuth in the browser when you first use it.
How to add the Linear MCP (Path B)
In Cursor (same as Atlassian, via the directory):
Open the Cursor MCP Directory.
In the search box, type Linear and find the Linear server (issue tracking, project management).
Click Add to Cursor. The install screen opens.
Click Install. Cursor adds the server.
Authorize: In a new chat, ask: “List my open issues” or “Create a story in project X.” A browser or prompt will open for Linear sign-in (or you will be asked for your API key). Complete sign-in. Or open Settings -> Cursor Settings -> Tools and MCP, find Linear and click Connect, then sign in when the browser or prompt appears.
Use it. Ask for issues, create stories, update status. The AI uses the Linear tools.
If Cursor asks to restart after install, restart once. Linear supports OAuth; if your setup asks for an API key instead, get it in Linear under Settings -> API -> Member API keys (or Security & access for personal keys) and paste it when asked.
In Claude Code (Connectors): Open Settings -> Connectors -> Browse Connectors, find Linear and click + to add it. Complete sign-in or enter your API key when prompted.
Create a basic Epic -> Story -> Task structure
After you connect, create one test hierarchy so the AI has a template for future requests. You do not need to click through the UI. Ask the AI:
Prompt to paste (Jira or Linear):
“Create a test hierarchy: one Epic named ‘Test: AI integration’, one Story under it (e.g. ‘First story’), and one Task or sub-task under that Story. Use my default project/team.”
The AI will create the Epic, then the Story under it, then the Task. If it asks for a project or team name, pick one you use. One example is enough; the AI will follow this structure when you ask for epics and stories later.
If you want a different structure (e.g. Project -> Milestone -> Story, or no Epics), describe it in a prompt:
“Create this structure: [describe levels and types]. One example of each level.” The AI will create what you described and reuse that pattern next time.
Over time you can refine the structure and fill in the fields that matter to you. With each prompt (create, update, search), your tool learns how you name things and how you like work organized. You stop opening Jira or Linear to hunt for the right issue or to click through filters. One prompt in chat finds the task, creates the story, or updates the status. That saves a lot of time once the habit is in place.
What you can do with Jira, Confluence, or Linear connected
Once the MCP is connected, you can ask in natural language. Some of the cases are listed below by category:
Create and structure
“Create an epic called [name] with description [text]”
“Create a story under epic [X]: [title and description]”
“Break this feature into 5 stories and create them in project [Y]”
“Create a task under story [Z]: [description]”
“Create a Confluence page in space [X] with title [Y] and link it to issue [Z]”
List and search
“List my open issues”
“Show issues assigned to me in project [X]”
“Find Jira issues with label ‘backend’ from the last 7 days”
“Search Confluence for pages about [topic]”
“What issues are in sprint [name]?”
“List all epics in project [X]”
Update and manage
“Move issue [ID] to In Progress”
“Set issue [ID] to Done”
“Add a comment to issue [ID]: [text]”
“Assign issue [ID] to [person]”
“Add label ‘urgent’ to issue [ID]”
Planning and breakdown
“Turn this PRD section into 3 stories and create them in [project]”
“What’s the status of epic [X]? List stories and their status”
“Create a story for [feature] with acceptance criteria in the description”
“Add sub-tasks to story [ID] for: design review, backend, frontend, QA”
Combined with email and calendar
“From my last email from [person] about [topic], create a story in [project]”
“I have a meeting with [person] tomorrow. Find open issues they’re assigned to”
“Create a Confluence page with meeting notes from today and link it to the relevant Jira epic”
Weekly and reporting
“Summarize what I completed this week (issues moved to Done)”
“List blockers: issues in progress for more than 3 days”
“What’s the status of epic [X]? Give me a short summary for the standup”
Takeaway
Add Atlassian or Linear MCP. One connection, one auth. After that, the AI can create epics and stories, list your backlog, update status, and search Confluence. Product work stays in your tools; the AI operates them from chat.
10-minute action
Pick your tool: Atlassian (Jira/Confluence) or Linear.
Add the MCP (steps above). Restart your app.
Complete sign-in (OAuth for Atlassian; for Linear, sign in when the browser opens or paste your API key when asked).
Use the prompt above to have the AI create one test Epic -> Story -> Task (or equivalent).
Try one use case: “List my open issues” or “Create a story: [title]”. If you get real data or a new issue, the MCP is working.
What to do next
Practice with the use cases above. Pick one that would simplify your day and speed up your work, and try it. For example: list your open issues, create a story from an email, or get a standup summary for an epic. Then think of your own use case, something you do often in Jira or Confluence that you wish was one prompt. Try it.
Reply with which tool you use (Jira, Linear, both). I’ll tailor the next prompts and use cases to your stack.
For paid subscribers: The examples above are a starting point. The full Prompt Pack (100+ copy-paste prompts) is a structured cheat sheet for Jira, Confluence, and your PMS. Use it to find the right prompt in seconds.
Create. Epics and stories from a PRD or email; issues, pages, comments; link issue to Confluence page; clone issue.
Read. Any issue, page, or comment; epic with stories; story with sub-tasks; changelog, linked issues.
Update. Status, assignee, labels, description, priority, due date; page content; bulk update multiple issues.
Delete. Comment, label, link; remove watcher; delete draft page.
List. My issues; by project, sprint, assignee, label, status; epics in project; stories in epic; blockers; overdue; no assignee.
Search. JQL, Confluence CQL; by keyword, label, assignee; full-text in space.
Link. Issues to each other (blocks, relates); page to issue; add story to epic; unlink.
Report. What I completed this week; blockers; standup summary; overdue list; by assignee; sprint summary; release readiness.
Plan. Turn feature into stories; add acceptance criteria; bulk-create sub-tasks; plan sprint; epic from theme.
Cross-tool. “Create a story from this email thread”; “List issues for the person I’m meeting tomorrow”; create page from meeting notes and link to epic.
One prompt does the job. No more digging in Jira or Confluence.
Keep reading with a 7-day free trial
Subscribe to Your Daily 1% Better at AI to keep reading this post and get 7 days of free access to the full post archives.