Plan sprints from Jira and Confluence
Sprint planning means reading specs in Confluence and re-typing them into Jira.
3 stepsModerate
Claude DesktopClaude CodeAtlassianSlack
How it connects
Your AI reaches your apps through small connectors called MCP servers.
Your AI
Claude DesktopClaude Code
Connectors
AtlassianSlack
Your apps
JiraSlack
This setup suits product managers and tech leads who write specs in Confluence and track work in Jira. Claude reads the spec page, proposes Jira issues with acceptance criteria and estimates, and creates them once you approve. You end up with a populated sprint backlog and a short summary posted to your team's Slack channel.
Set it up
3 steps. Open one for the details.
1Connect Atlassian
Add Atlassian's Remote MCP Server to Claude Desktop or Claude Code and sign in with your Atlassian account through OAuth. Claude can only see the Jira projects and Confluence spaces your own account can see, so check that you have access to the target project and the spec space. Run a test prompt such as listing the open issues in the project to confirm the connection.
Tip: The Atlassian server works with Atlassian Cloud sites; Jira or Confluence Data Center needs a different, self-hosted server.
2Draft the sprint
Paste the Confluence page link and ask Claude to break the spec into issues sized to fit a single sprint, each with a summary, description and acceptance criteria. Review the draft list in the chat first, merge or split items, and only then ask Claude to create them in Jira. Tell it which issue type, epic, labels and sprint to use so the issues land in the right place.
Tip: Ask for the draft as a table first; it is much easier to edit before anything is written to Jira.
3Share the plan
Connect the Slack MCP server and ask Claude to post a short summary of the sprint, with links to the new issues, to one named channel. Read the message preview before you approve the post. If your team prefers threads, ask Claude to reply in the existing planning thread instead.
Tip: Name the exact channel in your prompt so Claude does not guess.
Then just ask
Things to say once it's connected. Click one to copy it.
Ask Claude Desktop…Claude Desktop
Good to know
What you need, what it won't do, and how to keep it safe.
What you need
- Atlassian Cloud site with Jira and Confluence access
- Claude Desktop or Claude Code
- Slack workspace where you can install apps (for the summary step)
- A written spec in Confluence
Limits
- Claude does not know your team's capacity or velocity unless you provide it.
- Atlassian Data Center and Server installs are not supported by the hosted Atlassian server.
- Large specs with many linked pages may need to be split across several prompts.
Keep it safe
- Keep approval on for creating Jira issues.
- Limit Slack posting to one channel.
Questions
Straight answers before you start
Do I need a paid Atlassian plan?
No specific plan is required for the MCP server, but you need an Atlassian Cloud site with Jira and Confluence. Claude acts with your own permissions, so it cannot create issues in projects you cannot edit.
Can Claude estimate story points?
Claude can suggest estimates based on the spec, and it can write them into a story points field if your project has one. Treat the numbers as a starting point for the team discussion, since Claude does not know your team's velocity unless you give it past sprint data.
Which host should I use?
Claude Desktop works well for planning sessions because you can review drafts in a chat. Claude Code works too, and helps if you also want Claude to check the repository for related code.
What if Claude creates the wrong issues?
Keep tool approval on so you confirm each create call. If something slips through, you can ask Claude to update or close the issues, or fix them directly in Jira.
Tools in this setup
The products involved
