LabelSafe Rename, Merge & Undo Labels for Jira

Rename, merge and delete Jira labels safely: preview every change, run it in the background, and undo it in one click.

Jira has no rename or merge for labels, and deleting a label means editing every issue by hand. LabelSafe does it for you, safely. No external servers: data stays in Atlassian's Forge platform.

Key features

• Label list with approximate counts, site-wide or per project
• Duplicate scan: case, simple English plurals, short typos
• Rename one label, merge up to 20 into one, or delete a label
• Dry-run preview: JQL, approximate count, sample issues
• Background job: rate-limit aware, pauses at safety caps, resumable
• Activity log of the last 30 jobs, and one-click undo
• Project admins can clean their own project once a Jira admin allows it
• Bonus: a “Last comment” panel on the issue view

How it works

How it works. Pick labels, then choose rename, merge or delete. LabelSafe builds the query on the server, shows a preview, and then edits matching issues one at a time in the background. Jira has no rename-label API, so a label on thousands of issues takes thousands of edits. What the preview shows. An approximate, case-insensitive count, up to 20 scanned issues, the matching keys in that sample, and the JQL. It's a sample, not the full list of affected issues.

Limits

• Watchers may get emails. Every label edit asks Jira not to notify (notifyUsers=false), but that isn't a promise nobody is emailed: Jira honors the flag only when the app can administer the project, and Activity shows the outcome (requested, ignored, rejected or unknown). In every case, watchers may still be emailed.
• Counts are approximate and case-insensitive. "Bug" and "bug" can show the same count. Edits change only the exact label string.
• Safety caps: a job pauses at about 5,000 updated or 50,000 examined issues. Click Resume to continue.
• Undo reverses this job's label changes only. It doesn't restore a full snapshot of the labels field. A label someone else added later could be removed if this job had also added it. Deleted issues and edits Jira refuses are skipped.
• Issue security: issues the app can't edit are skipped and left out of undo, so the preview (what you can see) can differ from the job.
• Duplicate scan limits: it reads up to 1,500 label names and returns up to 40 groups, with English plurals only and a few irregulars. Short pairs like ui/ux are ignored.
• Label rules: no spaces, commas or control characters, 1–255 characters, and at most 20 source labels per operation.
• Last comment is an issue panel showing up to 280 characters. It's not an issue-navigator column or custom field.
• History kept: the newest 30 finished jobs. Older jobs, and their undo data, are deleted. A Jira administrator can delete job history from Activity → Clear history (type clear to confirm). It deletes all job records in the audit list, their undo pages and claims; it doesn't change labels on issues or settings, and it's refused while any job is queued, running or undoing. After clearing, those jobs can no longer be undone.

Permissions

Jira admins: Jira settings → Apps → LabelSafe. Project admins: Project settings → LabelSafe, only after a site admin turns on Allow project admins (it's off by default). The issue panel can't run label jobs.

Privacy

Undo data is issue keys plus label names. Summaries, descriptions and comment bodies are not stored. Logs carry only job id, status, error code and counts.

Docs

• FAQ and support
• Privacy policy
• Terms of use
• Coming soon to the Atlassian Marketplace