start

# Start Work on JIRA Ticket Begin work on JIRA ticket `$ARGUMENTS`. ## Step 0: Session Bootstrap ```bash bash .claude/scripts/session-setup.sh ``` Handles acli install+auth, recommended hooks, pre-commit, and gh auth check. Runs once per session (skips on subsequent calls). ## Step 1: View and Assign ```bash bash .claude/scripts/jira-ops.sh view $ARGUMENTS ``` Review the ticket summary, status, and description. Then assign and transition: ```bash bash .claude/scripts/jira-ops.sh assign $ARGUMENTS bash .claude/scripts/jira-ops.sh transition $ARGUMENTS "ASSIGNED" ``` ## Step 2: Check Backport Requirement ```bash bash .claude/scripts/jira-ops.sh check-version $ARGUMENTS ``` If Target Version is set, **backporting is required** after merge. ## Step 3: Create Branch ```bash DEFAULT_BRANCH="${PRIMARY_BRANCH:-$(git rev-parse --abbrev-ref origin/HEAD 2>/dev/null | sed 's|origin/||')}" DEFAULT_BRANCH="${DEFAULT_BRANCH:-master}" git checkout "$DEFAULT_BRANCH" && git pull origin "$DEFAULT_BRANCH" git checkout -b $ARGUMENTS-short-description ``` Branch naming: `<TICKET-KEY>-<kebab-case-description>`. Derive the description from the ticket summary. Ask the user if it's ambiguous. ## Step 4: Load Context Based on the ticket's area, read the relevant docs. The doc mapping is project-specific — check the project's `AGENTS.md` for the area-to-doc table. Common patterns: | Area | Doc | |------|-----| | API endpoints, auth | `agent_docs/api.md` | | Database, migrations | `agent_docs/database.md` | | Testing | `agent_docs/testing.md` | | Architecture | `agent_docs/architecture.md` | | Frontend | `web/AGENTS.md` or equivalent | ## Step 5: Report Summarize: - Ticket: key, summary, status, assignee - Backport: required or not - Branch: name created - Docs loaded - Next step: `/code`