BugForesight for Jira: how it works

BugForesight looks at the bugs your team has already filed in Jira. When someone opens a story, it shows the past bugs from the same area that the story doesn't seem to account for, and suggests a test scenario for each one. Every warning points to a real ticket on your site.

Installing

A Jira site admin installs BugForesight. Jira will mention that the app uses AI hosted by Atlassian; that's expected.

Right after install, BugForesight reads the issues of type Bug on your site and builds its own private history from them. On a small site this is quick. On a site with years of bugs it can take a few minutes. You don't need to do anything while it runs, and nothing on your issues changes, because the app only has read access to Jira.

Using it

Open any story and look in the Activity area of the issue. You'll find a tab called Bug Foresight.

The first time a story is checked, the tab says it's working. Once it finishes you get a list of past bugs, grouped by the kind of failure they had in common, with the most relevant ones at the top. Each one links to the original ticket, gives a short reason why it matters for this story, and offers a Given/When/Then scenario you can copy into your acceptance criteria or test plan. Edit the scenario to fit. It's a starting point, not a finished test.

Each warning has Useful and Not useful buttons. Your votes nudge how that warning ranks for you next time.

If you change the story and want a fresh look, press Check again.

AI checks and limits

One AI check means BugForesight reading one story for one person. Each Jira site gets a set number:

• Paid sites: 200 AI checks per calendar month, and up to 30 per day.
• Free trial (30 days): 10 AI checks in total, and up to 3 per day.

Daily counts reset at midnight UTC. Monthly counts reset on the 1st.

When a limit is reached, the tab keeps working. It shows keyword matches from your bug history instead, and says plainly that these haven't been checked against the story by AI. The same happens if the AI is slow (after about 90 seconds) or briefly unavailable. Keyword matches never run out.

BugForesight runs on Atlassian Forge and Jira Cloud, so Atlassian's own platform limits and rate limits apply too. A very large first-time history load can be slowed by them.

What it can and can't do

It's good at reminding a team about problems it has already seen. It works best when bugs have clear titles and descriptions, and labels or components that say which part of the product they belong to.

The warnings are AI-assisted suggestions. Sometimes they'll miss a related bug, and sometimes one will be off-topic. Treat them as a prompt for your team's review, not a verdict.

BugForesight never edits, comments on or creates Jira issues.

Your data

BugForesight runs on Atlassian Forge. We don't run any server that receives your Jira data, and nothing is sent outside Atlassian. The bug history, your team's votes and cached results are stored in a private database for your site inside Atlassian's app storage. The AI runs on models hosted by Atlassian.

The only personal data the app keeps is Atlassian account IDs, attached to votes and cached results. It doesn't store names, emails or avatars. When Atlassian tells us an account has been closed, BugForesight deletes that account's votes and cached results.

Full details: privacy policy and security and data handling.

Uninstalling

A site admin can uninstall BugForesight at any time from Jira's app management page. Your issues are untouched. The app's stored data is then removed by Atlassian under its Forge data retention rules.

Getting help

Write to support@beyondxia.xyz or see the support page.