Catch the weak ticketbefore the sprint does
A vague Jira ticket doesn't announce itself. It stalls three days in, when a developer can't tell what “done” means. Khint runs your quality bar against a ticket before it's pulled: score it on INVEST and your Definition of Ready, see the exact gaps, fix them, all from one keyboard shortcut.
“Looks fine” is not
a Definition of Ready
In planning, most tickets get a glance and a nod. The bad ones slip through: acceptance criteria you can't test, a title that says “improve the dashboard”, a story quietly three stories wide. You only find out mid-sprint, when the work stalls and the developer comes back with questions. A quality gate is supposed to catch this, but reading every ticket against the same checklist, every time, is exactly the dull discipline that erodes under deadline. Khint makes the bar a saved AI Action and runs it on demand, behind one palette shortcut, on whatever ticket you have selected.
One ticket, through the gate
Reviewing a teammate's story the morning before planning.
Write the bar down once, as an Action
Your Definition of Ready lives in everyone's head and nowhere useful. Turn it into a saved Action (“Ticket quality check”) that judges a ticket against INVEST and your team's checklist and returns a verdict plus the exact gaps. No prompt yet? Type the description and press Generate; refine it later with Improve. Now the standard is written down and runs the same way every time.
Pull the ticket you're about to pick up
With Atlassian connected, the palette's Jira Active sprint and Search rows surface the candidates for the next sprint. Use Pull issueand the ticket's current title, description, and acceptance criteria land in Khint, exactly as the author left them. How integrations work →
Run the check and read the verdict
Select the ticket text, hit Cmd+Shift+K, run your quality Action. Back comes the review: which INVEST letters it passes, where the acceptance criteria can't actually be tested, the “should be intuitive” line that means nothing to a developer, the missing rationale. A vague ticket fails out loud instead of stalling three days into the sprint.
Fix the gaps in place
Chain a second Action that rewrites the ticket to clear the flags, so “review → repaired ticket” is one keystroke. Or pair it with an acceptance-criteria Action that returns testable Given/When/Then for the criteria that failed. The improved version is pasted right back where the ticket already lives.
Put the review on the record
If the ticket belongs to someone else, the Jira Add comment row posts your review straight onto the issue so the author sees exactly what to fix, one issue per run, so every comment is deliberate. Keep a Memory session open for the planning prep and each check is logged, so by the end you have a record of which tickets passed the gate and which you sent back. How Memory works →
The rubric is yours to set
Khint applies the checklist you wrote. This is one example of what an Action might return.
There is no fixed score buried in Khint. The criteria above are just an example. The judgment lives entirely in the prompt you save, so a team that cares about testability, a team that cares about sizing, and a team with a long compliance checklist each get the review they asked for. Tighten the bar over time with Improve, and clone a starting point from the Backlog Refiner starter pack on the Agents page rather than writing from scratch.
Why not just ask a chatbot “is this ticket good?”
You can, and you'll re-explain your Definition of Ready every chat, get a different standard each time depending on how you phrased it, and copy the verdict back into Jira by hand. Khint pins the standard: the quality Action is written once and runs identically on every ticket, Memory carries the product context into the review automatically, and the Atlassian Jira rows let you pull the issue and post the review back without leaving the keyboard. Because an Action only sees the text you select (plus the session summary you opted into), the unreleased feature you're reviewing isn't sitting in a chat history.
Common questions
How does Khint score a Jira ticket's quality?
It runs the rubric you wrote. You save an AI Action that contains your quality bar (INVEST, your Definition of Ready, whatever your team agreed), and Khint applies it to the ticket text you selected. The result is whatever you asked the Action to return: a pass/fail per criterion, a short verdict, the specific lines that are vague or untestable. There is no fixed, hidden score inside Khint; the standard is the prompt you control, so two teams with different bars get different reviews from the same shortcut.
Is this the same as backlog refinement?
No. Refinement is the recurring pass where you take a stale backlog and make a batch of items ready. A quality check is a gate on a single ticket: is this one good enough to pull into the sprint, yes or no, and if not, exactly what is missing? You often run a quality check on a ticket someone else wrote, right before planning. Khint helps with both (see backlog refinement for the batch pass), but a quality Action is tuned to judge and flag one ticket, not to rewrite a whole list.
Can I review a ticket that's already in Jira?
Yes. Connect Atlassian with your email, an API token, and your cloud domain, and the palette gains Jira rows. Pull issue brings the ticket's current text into Khint, you run your quality Action on it, and if you want the verdict on the record you use Add comment to post the review back onto the issue, one issue per run, so nothing is changed silently. The active sprint and search rows help you find the tickets that still need a look before planning.
Do I have to write the quality prompt by hand?
No. In the Action editor, type a one-line description like "review this Jira ticket against INVEST and our Definition of Ready, list what fails and why" and press Generate. Khint drafts the prompt; press Improve later to tighten it in place. Khint also ships a Backlog Refiner starter pack you can browse and clone from the Agents page, which gives you a working ticket-quality Action to adapt to your own bar.
Where does the ticket text go?
An Action only ever sees the text you selected, plus the compacted summary of your active Memory session if you opted in. Memory sessions live in a local SQLite database on your Mac and are never synced to Khint's servers. Your Jira credentials sit in your OS keychain. Nothing about an unreleased ticket leaves your device silently, which matters when the thing you're reviewing describes an unannounced feature.
Put a gate on your next sprint
Free with 10 AI actions and 5 captures per day. No credit card. Write your bar once, check every ticket against it after.