Acceptance criteria,from a one-line story
A user story without sharp acceptance criteria is an argument waiting to happen in sprint planning. Khint turns a thin story into Given/When/Then scenarios, edge cases, and a Definition of Done, from one keyboard shortcut, in any app.
The story is easy,
the criteria are the work
“As a user I want to reset my password” takes ten seconds to write. What counts as done, which edge cases engineering hits on day one, and the scenarios that decide whether the ticket passes review are a craft you repeat story after story. Khint lets you encode that craft once as a reusable agent, then apply it to any story you select.
A workflow: the criteria are drafted, then filed where the team reads them, in one run.
One story, fully specified
Refining a password-reset story for the next sprint, step by step.
Save an acceptance-criteria agent once
On the Agents page, add an Agent to a pack: call it “Acceptance criteria”. Its prompt is yours to shape: turn the selected story into Given/When/Then scenarios, surface the edge cases that weren't spelled out, and finish with a Definition of Done. Don't want to write the prompt from scratch? Hit Generate to draft it from a one-line description, or browse the Backlog Refiner starter pack and clone it.
Start a Memory session for the feature
Refining a whole epic, not one story? Start a Memory session: “Password reset refinement”. Everything you run is logged to it, and Khint keeps a compact running summary that rides along with each AI call. So when you generate criteria for story #4, the agent already knows the constraints you settled in stories #1–3, without you re-pasting them.
Select the story, run the agent
Your story sits in Jira, a doc, or a notes app: “As a user I want to reset my password so I can get back into my account.” Thin on detail. Select it, hit Cmd+Shift+K, pick your Acceptance criteria agent, and the scenarios, edge cases (expired link, rate limiting, unknown email) and Definition of Done paste straight back where your cursor is. It works in any app, because Khint acts on the text you select.
Pull a story off a screenshot when you need to
Sometimes the story only exists as a design ticket or a spec image. From the palette, pick Extract text, drag a region over it, and the text is OCR'd to your clipboard. Paste, select, and run the same agent: the criteria come out the same way, whether the source was a text box or a picture. See how Capture works.
File the criteria where the team reads them
With Atlassian or Linear connected, the palette grows write rows: append the generated criteria as a Jira comment, spin up a new issue, or drop a Confluence page, without leaving your editor. Or chain it: a workflow can run your acceptance-criteria agent and then a Jira step in series, so a half-written story becomes a fully-specified ticket in one run.
Draft and file, in one run
Up to five steps in series: AI agents, then an integration write at the end.
The write step is what turns refinement into a filed ticket instead of a clipboard full of good intentions. Swap the drafting agents for your own house format, or point the last step at Linear or Confluence instead — the same workflow machinery writes to all of them, and the issue key comes back as the workflow's result.
Why a saved agent beats a chatbot tab
You could paste each story into a chatbot, and re-type your house style for criteria every time, re-explain the feature, and copy each result back by hand. A Khint agent holds your format permanently, so story #1 and story #40 come out consistent. The Memory session carries the feature context into every prompt automatically, the Atlassian and Linear integrations file the result where your team reads it, and because an agent only sees the text you select (plus the session summary you opted into), your backlog isn't accumulating in a chat history.
Common questions
What format does the acceptance criteria come out in?
Whatever your agent's prompt asks for: Khint doesn't hardcode a format. Save an agent that says "rewrite this story's acceptance criteria as Given/When/Then scenarios, list the edge cases I missed, and end with a Definition of Done", and that's the shape you get every time. Prefer a plain checklist or numbered list instead? Change one line in the prompt. If you're not sure how to word it, the agent editor's Generate button drafts the prompt from a short description, and Improve refines an existing one.
Can it read a story off a mockup or a screenshot?
Yes, in two honest steps. Khint's Capture pillar pulls the text out first: open the palette, pick Extract text, drag a region over the design ticket or the spec screenshot, and the raw text lands in your clipboard. Generating the criteria is then a second, explicit step: paste the text, select it, and run your acceptance-criteria agent. Capture extracts; agents structure.
Does Khint write the criteria back into Jira or Linear?
Yes, once you connect Atlassian (email, API token, cloud domain) or Linear. The palette then shows write rows: Jira Create issue and Add comment, Confluence Create page, Linear Create issue, so you can append the generated criteria to the story without leaving your editor. You can also pin a Jira create-issue or add-comment step inside a workflow, so "draft the criteria, then file them" runs in one shortcut.
Where does my backlog text go?
An agent only sees the text you selected, plus, if you opted in, the compacted summary of your active Memory session. Memory sessions live in a local SQLite database on your Mac or PC and are never synced to Khint's servers. Khint runs free at 300 credits a month, about 10 AI actions a day with no credit card; paid plans start at €7/mo, with Pro at €29/mo for 4,000 credits a month; every plan includes all the work integrations.
Try it on your thinnest backlog item
Free with 300 credits a month, about 10 AI actions a day. No credit card. Save the agent, select the story, watch the criteria fill in.