Goal
Give the team an on-demand way to check that Jira ticket work is complete on the feature branch before anyone opens a pull request.
Description
Semantic Scout is the ticket delivery-truth layer for Insights BI work. It reads the Jira summary, description, and comments, infers which artifacts belong to the ticket (dashboard, paginated report, semantic model), compares acceptance criteria to the git diff, scores Jira coverage and branch hygiene, lists missings, and surfaces likely review or bot feedback — so you fix gaps in Cursor instead of after a PR bounce.
Success looks like
Any AN or PIC ticket can be reviewed on demand; every acceptance criterion gets a clear met or missing line tied to files; the review card shows scores and READY before the PR; setup is once via Setup Semantic Scout; daily flow is Semantic Scout for AN-XXXX with no per-ticket checklists or one-off scripts on the branch.
What we can achieve
Standardized pre-PR review instead of scattered “are we good?” threads; faster intake when a new ticket starts; repeatable scope discipline on ticket branches; earlier catch of missing companions, documentation, or parity items reviewers usually ask for.
What we can use it for
Pre-PR gate on every dashboard, paginated, or model change; regression check after DAX, RDL, filter, or documentation edits; new ticket kickoff to confirm scope and expected files; self-service review when reviewers are busy.
What you will see
After you run it on a ticket, this is what each part means.
| On the screen | In simple words |
|---|---|
| Jira score | How much of the ticket is reflected in your changed files. |
| Branch score | How clean the branch is for review (right files, no stray folders). |
| Missings | What Jira asked for that is not in your branch yet. |
| Suggestions | Typical follow-ups: documentation, dashboard vs paginated parity, things reviewers often ask for. |
| READY | Scout’s summary call. If it says no, fix missings and run again before the PR. |
Who this is for
Insights / Glofox analysts on AN or PIC tickets — dashboards, paginated reports, semantic models.
Before vs after
Without Semantic Scout
PRs bounce for missing AC, extra files, or forgotten paginated or doc updates. Review goes around twice.
With Semantic Scout
One review card before the PR. You see scores and missings, fix them, then raise.
About 15 to 30 minutes back, each ticket
How it works
Setup — once
- Open Cursor.
- Type Setup Semantic Scout (install deps if Cursor asks).
- Connect Jira and GitHub once if prompted.
- Run Semantic Scout for AN-XXXX on a real ticket to confirm you see the card.
Daily use
Before every PR: Semantic Scout for AN-XXXX (or PIC-XXXX) on your feature branch. Fix what is missing. Run again if needed. Then open the PR.
Review card (what you get)
Live results come from your Jira ticket and git branch. Layout below is the standard format.
If this happens
| If you see | Do this |
|---|---|
| READY: NO | Fix the missings listed. Run Scout again. Do not open the PR yet. |
| Low Jira score | Compare each missing line to Jira AC. Update the right report or model file. |
If something fails
| What you see | What to do |
|---|---|
| Setup not found | Say Setup Semantic Scout again. Restart Cursor once. |
| No Jira text | Sign in to Jira in Cursor, or paste the ticket description in chat. |
| Wrong repo or branch | Open the product repo and checkout the branch that contains your ticket key; run Scout again. |