Bug Report to Engineering Issue
Example prompt: "When a support ticket gets tagged as a bug, summarise the report, check Linear for any existing issue about the same problem, and either add the new ticket as a comment on the duplicate or create a fresh issue with reproduction steps. Post the result to #bug-triage on Slack."
How to automate bug report triage with GloriaMundo
The Problem
Most bugs reach engineering through a clumsy chain of forwarded screenshots and half-paraphrased Slack messages. Support agents are not trained issue writers, and engineers are not trained ticket readers, so important detail goes missing in both directions. The same bug gets filed three times because nobody checked the existing backlog, and the engineering team spends as much time deduplicating as they do fixing.
How GloriaMundo Solves It
We build a workflow that triggers when a support ticket is tagged as a bug. An LLM step extracts the structured detail engineers actually need — what the customer was trying to do, what happened instead, the browser or device, any error messages, the account ID. Another LLM step searches the engineering tracker for existing issues that look like the same problem, scoring candidates on similarity. A conditional step splits the flow: if there is a strong match, the new ticket is added as a comment on the existing issue with the affected customer count incremented; if not, a new issue is created with the structured reproduction notes, a link back to the support ticket, and any attachments. A summary lands in the bug triage Slack channel so the team can see what was filed and what was merged. Glass Box preview shows you the candidate matches and the proposed issue draft before anything is written, so you can tune the similarity threshold and the issue template to match your team's style.
Example Workflow Steps
- Trigger (webhook): Fires when a support ticket is tagged with "bug" in the helpdesk.
- Step 1 (integration): Fetch the full ticket thread, customer metadata, and any attached screenshots from the helpdesk.
- Step 2 (LLM): Extract structured fields — expected behaviour, actual behaviour, environment, account, error text.
- Step 3 (integration): Search Linear for open issues with overlapping keywords from the extracted summary.
- Step 4 (LLM): Rank the candidate matches and decide whether any clears the duplicate threshold.
- Step 5 (conditional): If a duplicate is found, route to "merge"; otherwise, route to "create new".
- Step 6a (integration): For duplicates, add a comment to the existing Linear issue with the new customer's details and a link to the support ticket.
- Step 6b (integration): For new bugs, create a Linear issue using the structured reproduction template, attach the screenshots, and link the support ticket.
- Step 7 (integration): Post a summary to #bug-triage on Slack showing whether the report was filed fresh or merged, with links to both the Linear issue and the support ticket.
Integrations Used
- Linear — searches the backlog for duplicates and creates or updates engineering issues
- Slack — posts the triage summary so engineering and support can see the result
- Zendesk — source of the original support ticket and customer context
Who This Is For
Support leads and engineering managers at product teams where bug reports arrive faster than the triage rotation can keep up, and where the same bug regularly gets filed under three different titles.
Time & Cost Saved
A careful manual triage — reading the ticket, writing a reproduction case, searching the tracker for duplicates, filing or merging — takes around 10-15 minutes per bug. For a team taking 20 bug-tagged tickets per week, that is 3-5 hours of triage work, often falling on a senior engineer who would rather be fixing things. This workflow gets the report into a consistent shape and catches most obvious duplicates automatically, leaving the human a quick review of the proposed issue. The workflow uses integration, LLM, and conditional steps, costing a small handful of credits per ticket.