The notes are drafted before they're filed
The first agent reads exactly the lines you selected and gives back a ticket shape: one summary line, a description, and the acceptance criteria as bullets. Nothing reaches Jira until that draft exists.
Refinement ends with a list of decisions and no tickets. Khint drafts the ticket with AI (summary, description, acceptance criteria) and files it in Jira from one keyboard shortcut, on the notes you already typed.
A good refinement produces a dozen clear decisions and an hour of transcription work afterwards. That hour is where momentum dies and where vague tickets get filed. Khint drafts from the notes you already typed and files the result without a browser tab.
No window to open, no tab to switch, nothing to paste twice. Khint reads the selection where it is and writes the answer back in the same place.
The real palette, running an agent on a selection and then a workflow that ends in the tracker.
A sprint-refinement meeting, step by step.
Before refinement starts, open Memory and start a session: “Sprint 14 refinement”. Every action you run from now on is logged and Khint keeps a compact running summary, so each ticket you draft already shares the meeting's context. See how Memory works.
On Khint's Integrations page, paste your Atlassian email, an API token and your cloud domain into the Atlassian card. From then on the palette shows a JIRA section, and any workflow can carry a Create Jira issue step.
Mid-meeting you type the gist into your notes: “Export to CSV is timing out on big boards. New ticket: add server-side pagination. AC: under 2s for 5k rows, keep column order, no UI change.” No formatting, no Jira fields, just what was decided.
Select those lines and launch your workflow from the palette (Cmd+Shift+K). The first agent rewrites the notes into a structured ticket — a one-line summary, a clear description, a bullet list of acceptance criteria — and the final step is Create Jira issue.
The created issue key is the workflow's output. Paste it straight into the meeting thread so everyone sees the ticket exists. The filing is logged into your active session too, so by the end of the hour your Memory session reads like a refinement log: what was decided, what you filed, where it landed.
The draft, the issue key, the rows connecting Jira added to your palette, and the workflow that ties the three together.
The first agent reads exactly the lines you selected and gives back a ticket shape: one summary line, a description, and the acceptance criteria as bullets. Nothing reaches Jira until that draft exists.
The last step hands back what it created, so the key is already in your hands to paste into the meeting thread.
Six rows appear under a JIRA heading. The write you use every day is on by default, the reads are opt-in, and each one is a checkbox on the Jira card.
A workflow is up to five steps in series. The drafting is agents, the last step is the integration, and each step's result becomes the next step's input.
Up to five steps in series: AI agents, then an integration write at the end.
Swap the drafting agents for your own: your team's story template, a definition-of-done check, a translation step for a bilingual team. Or skip the hand-assembly and describe the flow in plain language — the workflow generator builds the workflow, reusing agents you already have and slotting in the Jira step once the integration is connected. The same machinery writes to Confluence and Linear, so a sibling workflow can publish the meeting summary as a page right after it files the tickets.
For a one-off ticket you don't need a workflow at all. Connect Jira once and the palette grows its own section, run straight on the notes you selected.
See the Jira integrationFiling the ticket is one shortcut. These three are what makes it one.
Yes, as a step you trigger, never in the background. Connect Jira with your email, an API token, and your cloud domain on Khint's Integrations page. Then either use the Create issue row in the palette, or add a Create Jira issue step to a workflow and pin the project and issue type in the editor. When you run it on selected notes, the first line becomes the ticket summary and the rest becomes the description, and the new issue key comes back as the result.
An AI agent does the drafting first. You write an agent ("rewrite these notes as a Jira ticket: a one-line summary, a description, and a bullet list of acceptance criteria") and it runs on the selected text before the Jira step files the result. The agent editor can Generate that prompt from a plain-language description and Improve it later, so you're not writing prompt boilerplate by hand.
If a Memory session is active, yes. Start a session for the meeting and everything you run is logged and compacted into a short summary that rides along with each AI agent, so the ticket draft already knows which feature you were refining without you re-explaining it. Sessions are stored locally in SQLite on your machine and are never synced to the cloud.
Only the text you selected, plus (if a Memory session is active) the compacted session summary. Nothing else leaves your device silently. Your Jira email, API token, and cloud domain are kept in your OS keychain, and every issue Khint files is logged into your local session history so you can trace what was created and when.
Free with 300 credits a month, about 10 AI actions a day. No credit card. Type the decision, draft the ticket, get the issue key back.