GitHub commits not showing in Jira: a checklist

You pushed a commit that mentions an issue, and the issue’s Development panel stays empty. This page goes through the causes in the order worth checking them. Most of it applies whatever connector you use: the Development panel belongs to Jira, and the rules for matching an issue key are the same everywhere.

1. Is the issue key written the way Jira expects?

A commit is matched to an issue through its issue key, for example CHK-42.

• Write it in upper case in commit messages and pull request titles: CHK-42 Fix login. Lower-case keys are a common reason for a commit that never links.
• It must stand as a whole word: CHK-42: and (CHK-42) work, xCHK-42 does not.
• It must be the key of the right project. If the project key changed, older commits carry the old key.
• Branch names count too: a branch called feature/CHK-42-login links the branch to CHK-42; some connectors, Linkwright included, also link that branch’s commits.

2. Was the commit pushed before the repository was connected?

Connectors learn about new commits through webhooks, which GitHub sends at push time. A commit pushed before the repository was connected never produced a webhook for it. Whether it is caught up later depends on the connector’s backfill. Linkwright’s, for example, rereads the last 100 commits of the default branch, so an old commit on a feature branch stays unlinked. Pushing a new commit that mentions the key is the quickest way to tell a backfill gap from a broken connection.

3. Is Jira simply still indexing?

Jira does not show development information the instant it receives it. Measured on a real Jira Cloud site on 30 September 2026: a commit reached the connector in 5.7 s and was sent to Jira in 7.2 s, but only appeared in the issue’s Development panel after about 57 s. Wait a minute before concluding that something is broken.

4. Are you reading a cached count?

The small summary on the issue (“3 commits”) can lag behind the details. On our test site, the summary kept showing 8 commits for a while after the detail view already listed all 11. Open the details (click the count) before deciding what is missing.

5. Are there more than 100 commits on the issue?

Atlassian documents that the Development panel displays at most 100 commits per issue. On a long-running issue, older commits are not lost: they are simply not all displayed.

6. Can Jira see it at all? Ask it with JQL

JQL searches the same development information as the panel:

development[commits].all > 0

lists every issue that has at least one commit, and

development[pullrequests].open > 0

every issue with an open pull request. If your issue is missing from the first query, Jira never received a link for it: the problem is upstream, in the connector or in the key.

7. Does the connection say it is healthy, and is it?

The most frequent complaint in the negative reviews of the official connectors is exactly this: every indicator says connected, and nothing gets linked. When the steps above do not explain a missing commit, check what your connector can actually tell you: when it last received an event from each repository, and when it last read that repository successfully. If it cannot tell you either, you cannot tell a delay from a failure.

How Linkwright handles this

Linkwright for Jira was built around point 7. For every repository, it shows the time of the last webhook, the time of the last successful full read, and a plain state: Up to date, Out of date, Rate limited, Failing or Not authorized, with the reason. A Refresh button rereads a repository from GitHub on demand, and the app checks five minutes after sending that Jira really kept the links, and repairs the ones it dropped. It feeds Jira’s native Development panel, so JQL and Automation work as usual.

It is free for up to 10 users. See the documentation or write to saad@linkwright.io.