<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: 404JacobPNotFound</title>
    <description>The latest articles on DEV Community by 404JacobPNotFound (@404jacobpnotfound).</description>
    <link>https://dev.to/404jacobpnotfound</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4244242%2Fbe670faa-e6df-49e2-98db-da631fde70ec.png</url>
      <title>DEV Community: 404JacobPNotFound</title>
      <link>https://dev.to/404jacobpnotfound</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/404jacobpnotfound"/>
    <language>en</language>
    <item>
      <title>Claude Jira: Built for Tickets, Slow at Team Metrics</title>
      <dc:creator>404JacobPNotFound</dc:creator>
      <pubDate>Thu, 19 Nov 2026 09:50:30 +0000</pubDate>
      <link>https://dev.to/404jacobpnotfound/claude-jira-built-for-tickets-slow-at-team-metrics-4263</link>
      <guid>https://dev.to/404jacobpnotfound/claude-jira-built-for-tickets-slow-at-team-metrics-4263</guid>
      <description>&lt;p&gt;This was first published on the Infograph blog: &lt;a href="https://infograph.ai/blog/claude-jira" rel="noopener noreferrer"&gt;https://infograph.ai/blog/claude-jira&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Connect Claude to Jira with Atlassian's own connector and it can find tickets, read them, write new ones, comment, and move them across the board, all with your Jira permissions. Atlassian built those tools around single tickets, so work on one ticket or one epic is what this setup does best. It's one connector in Claude, or one command in Claude Code, Anthropic's coding tool (&lt;a href="https://infograph.ai/blog/claude-jira#connect-jira-to-claude-with-atlassians-mcp-server" rel="noopener noreferrer"&gt;setup below&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;Team numbers are harder. Take cycle time: how long a ticket takes from the day someone starts it to the day it's done. Every ticket shows the day it was created, but the day someone started it is only in the ticket's history, its log of every change. So "What was our median cycle time last quarter?" needs the history of every ticket you finished, and Atlassian's history tool reads one ticket at a time.&lt;/p&gt;

&lt;p&gt;The shortcut Claude can take is to count from the day each ticket was created to the day it was resolved. That's lead time, which includes the weeks a ticket sat in the backlog, so it's usually a bigger number.&lt;/p&gt;

&lt;p&gt;So use Claude for the ticket in front of you, and a dashboard built on every ticket's history for the team's numbers.&lt;/p&gt;

&lt;h2&gt;
  
  
  Claude or a dashboard for your Jira questions
&lt;/h2&gt;

&lt;p&gt;Most of what a lead asks of Jira falls into two piles. One is about a ticket or an epic: what's blocking PAY-388, or write these notes up as a story. The other is about the team: where do our tickets wait, and are we getting faster?&lt;/p&gt;

&lt;p&gt;Claude with Atlassian's connector fits the first pile. It works where you already are, in a chat or in Claude Code, and it writes and moves tickets as well as reading them. One ticket or one epic takes a few tool calls (each call is one action, like fetching a ticket).&lt;/p&gt;

&lt;p&gt;The second pile is where it slows down. A question about many tickets' histories takes a call per ticket. Claude also works each answer out from scratch, so it may pick different dates or tickets from one week to the next. And because it can write, it can move the wrong ticket as easily as the right one.&lt;/p&gt;

&lt;p&gt;Infograph, which is what we make, fits the second pile. Our Jira connection keeps every status change on every ticket, so cycle time and time in status are counted across all of them, the same way each week. You get a dashboard: a board the team can look at in standup, and a link you can send to leadership.&lt;/p&gt;

&lt;p&gt;Its limits: it can't change Jira, so it won't write or move a single ticket. Its numbers are only as fresh as the last sync, and it needs a one-time setup, including which of your statuses mean waiting.&lt;/p&gt;

&lt;p&gt;You can also have both: Claude can ask Infograph's synced Jira history directly (&lt;a href="https://infograph.ai/blog/claude-jira#ask-claude-about-your-team-through-infograph" rel="noopener noreferrer"&gt;how, below&lt;/a&gt;).&lt;/p&gt;

&lt;h2&gt;
  
  
  Connect Jira to Claude with Atlassian's MCP server
&lt;/h2&gt;

&lt;p&gt;Atlassian makes the connector. In Claude's directory it's &lt;a href="https://claude.com/marketplace/connectors/atlassian" rel="noopener noreferrer"&gt;Atlassian MCP&lt;/a&gt;, marked Anthropic verified, and it covers Jira, Confluence and other Atlassian apps. MCP (Model Context Protocol) is the standard that lets an AI app like Claude plug into other software. Atlassian runs the server, so there's nothing to install: it hands Claude a set of tools, each a single action such as "get this ticket" or "move it to Done". It works with Jira Cloud, the version Atlassian hosts for you.&lt;/p&gt;

&lt;p&gt;On a Team or Enterprise plan, your Owner (Claude's account admin) adds Atlassian for the company first: Organization settings, then Connectors, then Browse connectors, then Add to your team (&lt;a href="https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities" rel="noopener noreferrer"&gt;Anthropic&lt;/a&gt;). Then everyone connects their own account:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Find Atlassian in Claude's connectors&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;In Claude on the web or the desktop app, open Customize, then Connectors, and press +. Find Atlassian MCP and press Connect (or Install).&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Sign in with Atlassian&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Follow the prompts to sign in with the Atlassian account you use for Jira. Claude then gets your existing permissions and nothing more, &lt;a href="https://support.atlassian.com/security-and-access-policies/docs/understand-atlassian-mcp-server/" rel="noopener noreferrer"&gt;Atlassian says&lt;/a&gt;. The first connection at a company may need your Atlassian admin to approve it (&lt;a href="https://www.atlassian.com/partnerships/anthropic/guide" rel="noopener noreferrer"&gt;Atlassian&lt;/a&gt;).&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Start with a question that only reads&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;In a new chat, press +, choose Connectors and check Atlassian is switched on. Then ask for something small, like the open tickets in one project.&lt;/p&gt;

&lt;p&gt;In Claude Code, &lt;a href="https://support.atlassian.com/atlassian-ai-gateway/docs/get-started-with-the-atlassian-remote-mcp-server/" rel="noopener noreferrer"&gt;Atlassian's command&lt;/a&gt; adds the server, and &lt;code&gt;/mcp&lt;/code&gt; inside a session signs you in:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;claude mcp add --transport http atlassian \
  https://mcp.atlassian.com/v2/mcp
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Plenty of guides still give a &lt;code&gt;/v1/&lt;/code&gt; address. If you used one, remove that connector and add Atlassian MCP again from the directory, or in Claude Code run &lt;code&gt;claude mcp remove atlassian&lt;/code&gt; and the command above. Atlassian moves anything left on v1 to the v2 tools on 1 March 2027 (&lt;a href="https://support.atlassian.com/atlassian-ai-gateway/docs/how-to-upgrade-from-atlassian-mcp-v1-to-atlassian-mcp-v2/" rel="noopener noreferrer"&gt;Atlassian&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;Claude acts with your own Jira permissions, so it can create, edit and move whatever you can. Its tools for deleting tickets and changing projects stay off until your Atlassian admin turns them on. That admin can also stop it writing to Jira at all, in Atlassian Administration under Connectors, then Atlassian MCP server, then Permissions, then Edit details (&lt;a href="https://support.atlassian.com/security-and-access-policies/docs/configure-atlassian-mcp-server-permissions/" rel="noopener noreferrer"&gt;Atlassian&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;In a chat, Claude asks for your approval before it uses a tool. Only choose Allow always for tools that just read, like a search or fetching a ticket. Research, Claude's long multi-step mode, is the exception: it uses connector tools without asking, so Anthropic advises switching off the tools that write before you use it (&lt;a href="https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp" rel="noopener noreferrer"&gt;Anthropic&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;If you came here for Claude writing the code behind a ticket, that's a different product: &lt;a href="https://www.atlassian.com/blog/company-news/claude-agent-for-jira" rel="noopener noreferrer"&gt;Claude Agent for Jira&lt;/a&gt;. You assign it a ticket and it proposes the code change on GitHub. Atlassian says it needs Jira's Standard, Premium or Enterprise plan with Rovo, Atlassian's AI, switched on.&lt;/p&gt;

&lt;h2&gt;
  
  
  Ticket jobs to give Claude in Jira
&lt;/h2&gt;

&lt;p&gt;Each of these touches one ticket, one epic or one search, which is what Atlassian's tools are built around. Name the project (and the Jira site, if you use more than one) so the search stays small, and ask for a draft before anything is created or changed.&lt;/p&gt;

&lt;p&gt;Monday, the bugs that came in over the weekend:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Find bugs in PAY created since Friday that nobody has assigned. For each one, suggest a priority and a component from its description, and flag any that look like duplicates of open bugs. Don't change anything yet.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;After planning, the notes nobody wants to type up:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Turn these notes into stories in PAY, each with acceptance criteria. Show me the drafts before you create anything.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Projects often have required fields of their own. Atlassian's server has a tool that reads which fields each issue type needs, so ask Claude to check them before it drafts.&lt;/p&gt;

&lt;p&gt;Midweek, when someone asks how the checkout epic is going:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Summarise epic PAY-210 in two short paragraphs for a weekly update: what's done, what's in progress, what's blocked and what it's waiting on.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;And before a release:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;List the Done tickets in fix version 4.12 and write release notes for customers, grouped into new features and fixes. Leave out internal tasks.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Why last quarter's cycle time is hard for Claude
&lt;/h2&gt;

&lt;p&gt;In Jira terms, cycle time runs from In Progress to Done, and lead time from the day a ticket is created. Our guide to &lt;a href="https://infograph.ai/blog/jira-time-in-status" rel="noopener noreferrer"&gt;time in status in Jira&lt;/a&gt; covers the difference, and Jira's own reports and the other ways to get both.&lt;/p&gt;

&lt;p&gt;Here's what the median for last quarter asks of the tools Atlassian documents:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Find the tickets.&lt;/strong&gt; One search lists the tickets resolved last quarter.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Read each ticket's history&lt;/strong&gt; (Jira calls it the changelog). Atlassian's history tool, &lt;code&gt;listJiraIssueChangelogs&lt;/code&gt;, reads &lt;a href="https://support.atlassian.com/atlassian-ai-gateway/docs/supported-tools/" rel="noopener noreferrer"&gt;one ticket's history&lt;/a&gt; per call, and nothing on its Jira list reads many at once. Jira's own API can return histories with a search, but Atlassian's page doesn't say its Claude tools can.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Do the sums.&lt;/strong&gt; Find when each ticket first moved into In Progress and when it reached Done, decide what to do with tickets that were reopened, and take the median.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Atlassian &lt;a href="https://developer.atlassian.com/cloud/rovo-mcp/changelog/" rel="noopener noreferrer"&gt;changes this list often&lt;/a&gt;, so check its supported tools page before you build a habit on it.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn46ep3dzvkeurszgm5bi.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn46ep3dzvkeurszgm5bi.webp" alt="Two questions side by side with the Atlassian MCP tools each needs. One ticket: Summarise PAY-388 and what's blocking it needs getJiraIssue and listJiraIssueComments, 2 calls. The team: What was our median cycle time last quarter? needs searchJiraIssuesUsingJql for tickets resolved 1 July to 30 September, discover to find the history tool, then executeRead listJiraIssueChangelogs once for each ticket, PAY-301, PAY-302 and on, with extra pages for long histories. Below: skip the histories and only created and resolved dates are left, which gives lead time, not cycle time." width="799" height="488"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;What each question asks of Atlassian's Jira tools, drawn from its supported tools page.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;If your team finished 200 tickets last quarter, that's 200 history reads, plus extra pages for long histories, before any arithmetic starts. That's slow going for one chat, which makes the shortcut tempting: every ticket carries its created and resolved dates on the ticket itself, with no history to read. If Claude counts from those, the number is lead time, whatever the answer calls it.&lt;/p&gt;

&lt;p&gt;Atlassian's own skills for Claude, the written instructions Claude follows for a job, avoid reading many histories too. The release check &lt;a href="https://github.com/atlassian/atlassian-mcp-server/blob/main/skills/assess-release/SKILL.md" rel="noopener noreferrer"&gt;reads the history of at most two blocked tickets&lt;/a&gt;, to date when each became blocked, because "only a changelog gives that reliably". The &lt;a href="https://github.com/atlassian/atlassian-mcp-server/blob/main/skills/jira-sprint-dashboard/SKILL.md" rel="noopener noreferrer"&gt;sprint dashboard skill&lt;/a&gt; judges how long unfinished work has sat by each ticket's last update instead.&lt;/p&gt;

&lt;p&gt;You can still ask. Just ask how it got there, too: which dates it used, and how many tickets' histories it read. Try one sprint before a whole quarter, and check two tickets by hand in their History tab. In Claude Code, an engineer can have Claude write a script against Jira's REST API that pulls every history and does the sums. That's a fair route, and it leaves you with one more script someone owns and reruns each week.&lt;/p&gt;

&lt;h2&gt;
  
  
  Answer team questions from synced Jira history
&lt;/h2&gt;

&lt;p&gt;A team number needs every status change for every ticket, kept in one place and counted the same way each week. That's what our Jira connection is for. You sign in with Atlassian once, and the first sync brings every ticket anyone has touched in the last two years, each with its full history. From then on a question about the quarter is one query over all of it, and the answer is a dashboard you can come back to.&lt;/p&gt;

&lt;p&gt;We built one on sample data for Rotaloom, a made-up software company with three Jira projects. The &lt;a href="https://infograph.ai/blog/jira-dashboard" rel="noopener noreferrer"&gt;Jira dashboard showcase&lt;/a&gt; walks through it, and the &lt;a href="https://app.infograph.ai/artifacts/b82e147b-ecb9-43a4-9769-ca46f71876f7" rel="noopener noreferrer"&gt;dashboard itself&lt;/a&gt; opens without signing in. Here is the same question over the last 90 days:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0eyniq5vuoqsukgwtjtz.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F0eyniq5vuoqsukgwtjtz.webp" alt="The Cycle time, ticket by ticket panel from the sample Jira dashboard. 287 tickets finished in the last 90 days, each a dot placed by the day it finished and the days it took from start to Done, on a scale of 0 to 40 days. A grey dashed line marks the median, 8.1 days, and a solid red line the 85th percentile, 16.0 days, with 43 tickets above it. Dotted lines mark three releases, each labelled: MOB 3.20, MOB 3.21 and APP 4.0." width="800" height="453"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Sample data. Each dot is one finished ticket, timed from its own history: half of the 287 took under 8.1 days, and the slowest 43, about one in seven, took longer than 16.0 days.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Each dot runs from the ticket's first move into In Progress to its last move to Done. Through Atlassian's tools that would be 287 history reads; here it's one query over history that's already synced.&lt;/p&gt;

&lt;p&gt;To build your own, &lt;a href="https://infograph.ai/connections/jira" rel="noopener noreferrer"&gt;connect Jira in Infograph&lt;/a&gt;. The guide covers each step and the errors you might see, such as a site that isn't in the list or an app your admin has blocked.&lt;/p&gt;

&lt;p&gt;Then ask Infograph for it. In Jira, a status like In Review counts as In Progress, even though the ticket is waiting for someone. So change the waiting statuses in the prompt to the ones your team uses.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Using my Jira data, build a dashboard of where our tickets wait. Show median and 85th percentile cycle time from In Progress to Done over the last 90 days, the median days finished tickets spent in each status, and the open tickets that have been in their current status longer than most finished tickets took in total. Treat In Review and Ready for QA as waiting.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;You can add any of these to the same question:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Lead time beside it:&lt;/strong&gt; Show lead time, from created to resolved, next to cycle time, so we can see how long tickets wait before anyone starts them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;By project:&lt;/strong&gt; Let me filter everything by project and issue type.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Work in progress:&lt;/strong&gt; Chart how many tickets were in progress each week against that week's cycle time.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Bugs that bounce:&lt;/strong&gt; Show the bugs that went from Done back to In Progress, and how much longer they took.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;One Jira connection fits Infograph's Free plan, where you sync by pressing Sync. On the Plus plan it syncs daily, and on Pro every 30 minutes (&lt;a href="https://infograph.ai/#pricing" rel="noopener noreferrer"&gt;see the plans&lt;/a&gt;). Infograph reads with your Jira permissions and can't change anything in Jira.&lt;/p&gt;

&lt;h3&gt;
  
  
  Ask Claude about your team through Infograph
&lt;/h3&gt;

&lt;p&gt;To ask from Claude instead, add Infograph as a custom connector beside Atlassian's:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Add Infograph as a custom connector&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;In Claude, open Customize, then Connectors, and press + Add, then Add custom connector. Name it Infograph and paste &lt;code&gt;https://app.infograph.ai/mcp&lt;/code&gt;.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Sign in to Infograph&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Press Continue, keep the suggested sign-in settings, press Continue again, then Add, and sign in to Infograph.&lt;/p&gt;

&lt;p&gt;On a Team or Enterprise plan, your Owner adds it first, in Organization settings, then Connectors (&lt;a href="https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp" rel="noopener noreferrer"&gt;Anthropic&lt;/a&gt;). Claude's Free plan allows one custom connector.&lt;/p&gt;

&lt;p&gt;In Claude Code, it's one command, then &lt;code&gt;/mcp&lt;/code&gt; to sign in:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;claude mcp add --transport http infograph \
  https://app.infograph.ai/mcp
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Your Jira site has to be connected in Infograph first. After that, Claude can ask the synced history directly: Infograph runs the question over every ticket and sends back just the answer. Keep Atlassian's connector for the ticket work; the two sit side by side.&lt;/p&gt;

&lt;p&gt;A rule of thumb for the two: if the question has a ticket key in it, ask Claude with Atlassian's connector. If it has a date range in it, ask the synced history.&lt;/p&gt;

&lt;h2&gt;
  
  
  Claude Jira questions
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Is there an official Claude Jira connector?
&lt;/h3&gt;

&lt;p&gt;Yes. Atlassian MCP, built by Atlassian, is in Claude's connector directory and marked Anthropic verified. It covers Jira, Confluence and other Atlassian apps, and works with your existing Atlassian permissions.&lt;/p&gt;

&lt;h3&gt;
  
  
  What is the Jira MCP server URL?
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://mcp.atlassian.com/v2/mcp" rel="noopener noreferrer"&gt;https://mcp.atlassian.com/v2/mcp&lt;/a&gt;. Older guides give a v1 address. Atlassian says that from 1 March 2027 anything still on v1 starts getting the v2 tools; in Claude, replace an old v1 connector with the one in the directory, and in Claude Code sign in again.&lt;/p&gt;

&lt;h3&gt;
  
  
  Why is the Atlassian connector not working in Claude?
&lt;/h3&gt;

&lt;p&gt;Atlassian's troubleshooting page says an expired sign-in fails silently, so disconnect and reconnect it in Customize, then Connectors. A first connection at a company may need your Atlassian admin to approve it. And if your company only lets certain IP addresses reach Atlassian, Atlassian notes that calls from a hosted AI app can come from the AI provider's addresses rather than your VPN, so an admin may need to allow them.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can Claude create and update Jira tickets?
&lt;/h3&gt;

&lt;p&gt;Yes. Atlassian's tools let it create tickets, edit them, move them between statuses and comment, with your Jira permissions. Its tools for deleting tickets and changing projects are off until your Atlassian admin turns them on.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can I make Claude read-only in Jira?
&lt;/h3&gt;

&lt;p&gt;Your Atlassian admin can block writing for Jira in Atlassian Administration: Connectors, then Atlassian MCP server, then Permissions, then Edit details. On Claude's Team and Enterprise plans an Owner can also set each tool to Always allow, Needs approval or Blocked for everyone. On your own plan, read each approval and only choose Allow always for tools that just read.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can Claude calculate cycle time from Jira?
&lt;/h3&gt;

&lt;p&gt;It can, by reading each ticket's history. Through Atlassian's documented tools that's a history read per ticket, or an engineer can have Claude Code script it against Jira's REST API. Either way, ask which dates it used: created to resolved is lead time, not cycle time.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is there a Claude Jira skill?
&lt;/h3&gt;

&lt;p&gt;Atlassian publishes skills for Claude, written instructions it follows for a job, in its atlassian-mcp-server repository on GitHub: triage-issue, capture-tasks-from-meeting-notes, generate-status-report, assess-release and more. Atlassian lists them as one of its ways to connect Claude to Jira. Read any skill before you add it, since it can tell Claude to use every tool it has.&lt;/p&gt;

&lt;h3&gt;
  
  
  Does the Atlassian connector cost anything?
&lt;/h3&gt;

&lt;p&gt;Anthropic says connectors are available to all Claude users. On Atlassian's side, only tools marked Enriched on its supported tools page use Rovo credits, Atlassian's allowance for its AI features, and the Jira tools in this post aren't marked that way.&lt;/p&gt;

&lt;h3&gt;
  
  
  Does it work with Jira Data Center or Server?
&lt;/h3&gt;

&lt;p&gt;Atlassian lists an Atlassian Cloud site as the requirement for its MCP server. Infograph's Jira connection is also for Jira Cloud only.&lt;/p&gt;

&lt;h3&gt;
  
  
  Can I connect Jira to ChatGPT the same way?
&lt;/h3&gt;

&lt;p&gt;Atlassian's guide says to add its MCP server as a connector in ChatGPT and sign in with Atlassian, and it uses the same address. The tools are the same, so the same limits apply to team numbers. What you can add depends on your ChatGPT plan and workspace settings.&lt;/p&gt;

&lt;h3&gt;
  
  
  I'm having trouble connecting Jira to Infograph
&lt;/h3&gt;

&lt;p&gt;The guide covers each step and the errors you might see, such as a site that isn't in the list or an app your admin has blocked: &lt;a href="https://infograph.ai/connections/jira" rel="noopener noreferrer"&gt;Jira guide&lt;/a&gt;, including &lt;a href="https://infograph.ai/connections/jira#troubleshooting" rel="noopener noreferrer"&gt;troubleshooting&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Originally published at &lt;a href="https://infograph.ai/blog/claude-jira" rel="noopener noreferrer"&gt;https://infograph.ai/blog/claude-jira&lt;/a&gt;&lt;/p&gt;

</description>
      <category>jira</category>
      <category>claude</category>
      <category>ai</category>
      <category>mcp</category>
    </item>
  </channel>
</rss>
