The job
You’re in a Jira issue with acceptance criteria, edge-case notes, and comment-thread decisions, and you need a clean set of test cases before implementation starts. Instead of drafting from memory, you want unit, integration, and failure-path coverage generated from the ticket text you’re actually shipping against.
Why this is hard without Sephir
Without Sephir, you bounce between Jira and a separate chat window, manually copy text blocks, and lose important context hidden in comments or linked tabs. If the ticket references another page, you repeat the same copy-paste loop and re-explain the prompt each time. That workflow turns test design into clipboard maintenance, not engineering.
How Sephir does it
- Pin the Jira issue tab you want to convert into tests.
- Open the Sephir sidepanel with
Cmd+Shift+S. - Ask for structured unit, integration, and negative-path cases from the acceptance criteria.
- Watch the timeline run
extractPageText(tab)and return a typed test-case structure. - Export the output as Markdown or JSON for your repo, QA doc, or PR checklist.
- Save the successful trace as
/jira-testsand reuse it on the next ticket.
One slash command gives you the same structure every sprint.
The skill behind it
This skill uses Jira page text as the source of truth, then outputs consistent, schema-aware test cases so you can review, edit, and commit quickly without rewriting the prompt every time.
What it costs
Sephir runs this on your own provider setup: ChatGPT Plus via Codex OAuth or your API key. Typical usage is ~3,000–6,000 input tokens and ~500–1,200 output tokens per run on Claude Opus 4, GPT-5.5, or Gemini 3 Pro. Cost depends on your model choice; Sephir itself is detailed on .