What problem does it solve? Before writing an implementation plan, engineers need a complete picture of the problem: what the ticket actually says, what the relevant code does today, and where the change pressure is. Doing this manually means juggling ticket comments, linked MRs, and scattered code exploration, often producing shallow or duplicated context. ## Core Features & Use Cases - Ticket deep-loading: A dedicated subagent reads the full ticket, all comments, and transitively linked MRs, design docs, and chat threads, dumping findings into a markdown file without touching the codebase. - Two-stage code exploration: A surface-level map of affected subdomains is built first, then focused per-subdomain deep dives run in parallel or sequence depending on overlap. - Structured briefing output: Synthesizes problem, current state, change pressure, open questions, and suggested directions into ./.claude/plans/pre-plan-<slug>.md, optionally with Mermaid diagrams. - Use Case: Given a Jira ticket ID, the skill loads the ticket and its linked MRs, maps the three subsystems it touches, deep-dives each one, and produces a context briefing ready for a solution discussion with the user. ## Quick Start Ask the AI to run the pre-plan skill with a ticket reference or a freeform problem description to gather context before planning.
