jira-contextlisted

# Jira Context Turns a git branch into the **requirement context behind it**: finds the Jira key, fetches the ticket, and extracts acceptance criteria + reproduction steps as concrete, traceable inputs for QA. Standalone, and also used by `test-iteration` (step 1). Read-only — it never writes to Jira. ## 1. Resolve the ticket key (deterministic) Run the bundled script — do not guess the key: ```bash "${CLAUDE_PLUGIN_ROOT}/scripts/jira-key.sh" <branch> [base] ``` It extracts a key in the official Jira format (`[A-Z][A-Z]+-[0-9]+`) from, in priority order: **branch name → PR title (`gh`) → non-merge commit messages**, and prints `key:` + `source:`. If `${CLAUDE_PLUGIN_ROOT}` is unset (manual install) or the script is missing, run the equivalent inline: `printf '%s' "<branch>" | grep -oE '[A-Z][A-Z]+-[0-9]+' | head -n1`. **Done when:** you have a key + its source, or `NONE`. ## 2. Fetch the ticket — tool-agnostic with fallback - If a key was found **and** the Atlassian MCP is available, fetch it. The MCP tools may be deferred — first `ToolSearch query:"select:mcp__plugin_atlassian_atlassian__getJiraIssue"` to load the schema, then call `getJiraIssue` for the key. Request the fields QA needs: `summary, issuetype, description, priority, components, labels, status, issuelinks`, plus whether attachments exist. Skip worklog, watchers, sprint metadata, estimates — noise for test design. - **Pull the signal comments** (not the whole noisy thread). The discussion carries what the