One story too bigbecomes three that ship
Every backlog has the story nobody can estimate. Khint pulls it from Jira into a Memory session, splits it into thin vertical slices with a reusable AI action, and files each slice back, all from one keyboard shortcut.
The eight-pointer
that rolls over every sprint
You know the story: too big to estimate, too tangled to start, carried from sprint to sprint until nobody remembers why it was written. Splitting it well is a craft, thin vertical slices that each deliver value, and doing that craft by hand for every oversized ticket is the part that never gets budgeted. Khint turns your favorite splitting patterns into a saved Action you run in seconds, with Memory carrying the ticket and its constraints into every prompt. Starting further upstream, with a whole epic instead of one story? That is a different job: see the epic breakdown walkthrough →
One oversized story, start to filed slices
Splitting a “customer can manage payment methods” story that would not fit the sprint, step by step.
Start a refinement session
Before you open the board, start a Memory session, call it “Checkout refactor refinement”. Every pull and every Action you run from now on is logged into it, and Khint keeps a compact running summary that rides along on each AI call, so the split stays anchored to this story and its constraints. How Memory works →
Pull the oversized story from Jira
With Atlassian connected, run the Jira Pull issue row from the palette: pick the story from a list of your recent issues, or type its key. The ticket's summary, status, and description land as an entry in your active session. No copy-pasting the story into a chat window, and no retyping the background. How integrations work →
Save a reusable splitting Action
In Agents, create an Action like “Split story into thin slices”. No prompt yet? Type a one-line description and press Generate: Khint drafts a prompt that asks for two to four vertical slices, each independently valuable and small enough to estimate, with the pattern used (workflow step, business rule, happy path first) named for each slice. Or clone the Backlog Refiner starter pack and adapt a working Action.
Run the split where the story lives
Select the story text, in Jira in your browser or in your planning doc, hit Cmd+Shift+K, and run the splitting Action. The candidate slices are pasted back in place. Because your session context is attached, the model already has the pulled ticket and whatever constraints came up earlier in refinement, so the slices respect them instead of contradicting them.
File the slices, keep the debate on the ticket
Two shipped paths back to Jira. Select a slice and run Create issue to file it under your pinned project and issue type, one issue per run. Or run Add comment on the original story to post the proposed split where the team will see it, and let refinement decide before anything new is created. Each slice that needs sharper criteria gets a second pass with an acceptance criteria Action →
Why not just paste the story into a chatbot?
You can, and you'll restate your slicing rules in every new chat, paste the ticket and its comments in by hand, and copy each slice back out one by one. Khint removes the round-trips: the splitting Action is saved once and reused on every oversized story, the Jira pull lands the ticket in your session so Memoryfeeds it to the model automatically, and Create issue and Add comment put the result back where the team actually reads it. Because Actions only see the text you select, plus the session summary you opted into, your backlog isn't sitting in a chat history.
Common questions
How does Khint split a user story?
You save a reusable Action (an AI prompt) that takes one oversized story and returns two to four thin vertical slices, each one demoable and shippable on its own. The prompt encodes the splitting patterns you trust: by workflow step, by business rule, happy path first with edge cases later. Select the story text anywhere, hit Cmd+Shift+K, run the Action, and the split is pasted back in place. Khint sends only the text you selected, plus your active session summary if you opted in.
Can Khint read the story from Jira first?
Yes. Connect Atlassian with your email, API token, and cloud domain, and the palette gains a Jira Pull issue row: pick from a list of your recent issues or type any key. The ticket's summary, status, and description land as an entry in your active Memory session, so every Action you run afterwards already knows what the story says. Pulls require an active session, because landing the ticket in the session is the whole point. Each Jira row can be switched on or off from the MCP page.
How do the slices get back into Jira?
Two ways, both from the same palette. Jira Create issue files one slice per run under the project and issue type you pinned, so each new story is created explicitly. Or use Jira Add comment to post the proposed split on the original oversized story, so the team can debate the slicing in refinement before anything is created. Both flows show a preview behind a confirm button before writing.
Is this different from breaking down an epic?
Yes, and Khint has a separate walkthrough for that. Epic breakdown starts upstream: one large intent decomposed into a first set of stories. Story splitting starts later, in refinement, when a single story turns out too big to estimate or finish in a sprint and needs to become thinner slices without losing its value. Same palette, same reusable Actions, different prompt and different moment in the process.
Try it on the story your team keeps carrying over
Free with 10 AI actions and 5 captures per day. No credit card. Save the splitting Action once, slice every oversized story after.