goal-wraplisted

# /goal-wrap — Grounded, tracked goal bundle ## Host dispatch contract - Codex: `$skill`, Codex collaboration roles, and GPT-5.6 tiers. - Claude: `/skill`, Agent/Skill tools, and Claude aliases. - A bare `/skill` in this shared source denotes the Claude form; Codex dispatches the same named skill as `$skill`. Pack current conversation state into a self-contained execution bundle: 1. **Best-effort research first** — a small research pass grounds the goal in real architecture before handoff (codebase search + prior-decision recall, using whatever tools are actually available — see "Soft dependencies" below) 2. **Tracked DONE WHEN** — each criterion has a proof command; executor runs it and updates status; goal clears only when all show `[✓]` 3. **Full autonomy by default** — commits, push, merge, DB migrations, prod deploys, schema changes all pre-approved; pass `--gates` to revert to ask-first behavior 4. **Parallel-execution hint baked in** — goal instructs the executor to use this package's own `/feature-implement` (or a manual swarm) for parallel work Use before: `/clear`, agent handoff, `/loop`, `/schedule`, switching machines. ## Soft dependencies (degrade gracefully — do not block on absence) This skill was originally authored inside a monorepo with a codebase-index MCP server (`repowise`), a cross-session memory store (`gbrain`), and two sibling skills (`/prompt-master`, `/handoff`). None of those are guaranteed to exist for a standalone `feature-fi