User Story

As a Chief-of-Staff or tech lead kicking off a piece of work that decomposes into 5–20 tickets, I want a skill that files them all in one flow — without the slow interactive path (/feature × 20 = ~100 turns) or the non-conformant fast path (raw gh issue create with whatever shape the agent invents).

Summary

New skill /tickets-batch at .claude/skills/tickets-batch/SKILL.md. Bulk-files a numbered/bulleted list (or markdown file) of ticket titles by:

1. Resolving the repo (session marker → registry → ask).
2. Asking SHARED-CONTEXT questions once for the whole batch: default priority, optional epic, optional area labels, repo confirmation.
3. Per ticket, a 3-question micro-interview: type (Feature/Bug/Chore/Refactor/Testing/CI/Docs), one-line purpose, optional clarification when the inferred body is low-confidence.
4. Showing the full batch summary table for confirmation.
5. Filing each via gh issue create — output conforms to .ticket.required_sections (from [Chore] Make ticket-prefix whitelist + schema project-configurable (not hardcoded) #109 + [Chore] validate-issue-structure.sh — config-driven PreToolUse hook for gh issue create #107) by construction.
6. On first failure: stop, surface the validator error, ask retry/skip/edit/abort.

The skill replaces the bypass scenario where an agent files 10 tickets via raw gh issue create with bespoke bodies (the failure mode that triggered apexyard#107 + the related skill-discipline memory). Now the happy path matches the team standard.

Schema source: .ticket.prefix_whitelist, .ticket.label_priority_scheme, and .ticket.required_sections from project config — no hardcoding, so future schema rolls forward without touching the skill.

Testing

Skills are markdown specs — no shell test fixture. Behaviour is exercised by the agent following the spec. Conformance is enforced at file time by validate-issue-structure.sh (#107) — the skill produces output that passes that hook.

Smoke check on conformance:

# Spec must reference required_sections per type — verified by inspection
grep '## Required body shape' .claude/skills/tickets-batch/SKILL.md
# Spec must call gh issue create per-ticket (not batch-API), so the validator runs per-issue
grep 'gh issue create' .claude/skills/tickets-batch/SKILL.md

Real dogfood will be the next batch-file moment (e.g. a project kickoff with 5+ tickets) — the spec is followable as-is, and any rough edges surface there.

Scope — what this does NOT do

• No automatic detection of "this is a bulk intent" from the user's prompt — the agent decides when to invoke /tickets-batch vs. /feature (covered by the parallel-skill-invocation rule from [Feature] /fan-out skill — spawn N parallel agents on independent tasks, with per-task agent type + worktree isolation #117).
• No bulk-edit flow — creation only.
• No per-ticket priority override (priority set once for the whole batch, re-prioritise after if needed).
• No integration with non-GitHub trackers (Linear/Jira). Existing skills are GitHub-only; this matches.

Glossary

Closes #108