Your agents don’t need to persist. Your purpose does.
How do I connect Jira to Claude?
This archived article remains available at its original address. Unl, your why agent, carries the decisions you settle and why into your AI work. See how it works.
Atlassian's MCP surface covers Jira and Confluence together — issues, sprints and status alongside the docs — which is usually how the questions actually arrive.
Direct answer: connect the Atlassian MCP surface in Claude (Settings → Connectors → authorise your site; 'claude mcp add' in Claude Code) and Claude reads Jira issues, sprints and status, with Confluence docs alongside. Sprint questions, ticket archaeology and cross-referencing a spec against its epic all become conversational reads.
The backlog answers questions now
What entered the sprint late, which epics carry the oldest open bugs, what changed on the ticket since Thursday, what the linked Confluence spec actually commits to — pulled live rather than reconstructed from memory and tabs.
A threshold Jira displays but never enforces
Worked case: your team ratified a bug-debt gate — when open customer-reported bugs pass twenty-five, feature work pauses until the count is back under — set after a support quarter nobody wants repeated. Jira counts the bugs perfectly. The pause is a promise you made each other, and promises live outside the tracker.
The tracker read with the promise attached
Through Unl the sprint read arrives holding the gate: twenty-eight open — over your line, feature work pauses by the rule you set last autumn. Planning starts from the standing agreement rather than from someone remembering to raise it. That is the difference between a threshold on a dashboard and a threshold with authority.
Jira to Claude reads the sprint; Jira through Unl reads the sprint against the agreements you ratified about it — the gate fires because you set it, and the reason is right there.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. It works today: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
What is on the Atlassian MCP read surface?
Jira issues, sprints and status, and Confluence docs — the pairing matters, since half of real Jira questions end in 'and what did the spec say?'
Can ChatGPT use the same connection?
Yes; the surface is MCP-standard, so 'connect Jira to ChatGPT' is the same setup in a different client menu. The measured layer through Unl equally reaches any MCP surface.
What is a good first rule to ratify for a Jira-heavy team?
Pick the agreement you re-explain most: a bug-debt gate, a WIP limit per engineer, a definition of blocked. Ratify one, with its incident, and let the first framed sprint read make the case for the rest.
What if the rule fires on nearly every ticket?
Then either the rule is wrong or the work is, and the reason attached to it is what tells you which. A gate that fires constantly gets ignored within a fortnight, so treat a high hit rate as a finding about the rule rather than as evidence the team is failing.
What this is
Your agents don’t need to persist. Your purpose does.
Save the thoughts, decisions and targets worth keeping, each with its reasoning, carried into every AI session the moment they matter. A new unit of exchange between you and your AI: the Settled Why with standing that travels. Unprompted.
Your whole AI world. What you decided at the epicentre. It works inside Claude, Claude Code, ChatGPT and Cursor, and connects in a couple of steps.
MCP native·Human settled·Model agnostic·Your data
Measured Context
Connect Unl to bring the right information into the moment.
Your sources, read against the criteria you set.
The full product, open. Free to start. Paid plans from $10 a month.