ticket-fetchlisted
Install: claude install-skill YevheniiaQA/qa-agent-toolkit
# Ticket Fetch
The pipeline's entry point. Everything downstream is only as good as what this
skill recovers, so recover everything — not just the description field.
## Disabled integrations
Follow `qa-conventions` → *Disabled integrations* for the tracker or the wiki (`jira`, `linear`, `confluence`): read `INTEGRATIONS`, skip a
system that is not listed, report it as `➖ <system> disabled for this project`, never route
around a `qa-guard` block.
**Local alternative:** take the requirements from the markdown files the engineer names, or from the workspace when the profile says `TICKET_SOURCE=local`. Write `_source.json` beside them (path, who supplied it, date) instead of `_fetch.json`: the provenance gate in requirements-decomposer accepts either, and a synthesised ticket passes neither.
## What to collect
1. **The ticket itself** — summary, description, type, status, priority, labels,
components, fix version, and any custom fields the team actually uses.
2. **Sub-tasks, recursively.** Requirements are frequently split across a parent
story and its backend/frontend children. Fetch the whole subtree.
3. **Attachments, parsed.** Download and extract text from PDF, DOCX, XLSX, PPTX.
A requirements spreadsheet listed as a filename is useless; its contents are not.
4. **Linked Confluence pages**, including child pages one level deep.
5. **Comments.** Requirements get amended in comments far more often than teams
admit. Capture them with author and timestamp so la