# Whats Next

> Re-orient the user after they return to a session — or after they open a FRESH context pointed at a worktree/branch — and let them decide the next step by answering lightweight questions through the host's structured-question tool (Yes/No, A/B/C/D) instead of facing a blank prompt. Trigger when the user invokes /whats-next (optionally naming a worktree, branch, or repo path), or asks "接下来做什么", "下一步是什么", "我该做什么", "这个 worktree/分支做到哪了、下一步呢", "what's next", "what should we do next", "帮我决定下一步", or returns after a long gap and wants to resume work with minimal typing. Works with zero conversation history: state is then rebuilt from git evidence (diffs, branch commits, uncommitted changes) and handoff/task docs. NOT for pure recap with no decision needed (that is a recap request — answer it directly or use a context-rebuild skill if available), and NOT for questions about a specific file or piece of code.

- Skill: `luw2007/whats-next` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add luw2007/whats-next`
- Raw SKILL.md: https://api.skillmd.com/api/skills/luw2007/whats-next/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: luw2007 (https://skillmd.com/u/luw2007)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/luw2007/whats-next

---


# What's Next

The user runs many sessions in parallel and comes back after hours away. They cannot
remember where this conversation stands. The job: rebuild the state FOR them, then hand
them a small set of concrete, clickable decisions — never a blank prompt to fill in.

Match the language of the conversation.

## Required reference

Read [references/workflow.md](references/workflow.md) before acting — it holds the full
step-by-step (evidence source → state reconstruction → recap → fork points → optional
external context → structured question → act) plus edge cases (nothing pending, external
waits, cold-start worktree, substantive user-only decisions).

## Workflow overview

1. **Pick evidence** — warm session: use conversation + a cheap `git status`/diff check.
   Cold start / named worktree/branch: rebuild entirely from git evidence and handoff docs.
2. **Reconstruct state silently** — DONE / IN FLIGHT / BLOCKED / PLANNED, each graded by
   evidence strength; never fabricate status for anything still running.
3. **Recap in 3–5 lines**, always leading with an explicit close-session verdict.
4. **Identify real fork points** — decisions that actually change what happens next, not
   already-decided or fake choices.
5. **Optionally enrich with external context** (PR/CI, issue tracker, running work) only
   when it would change an existing fork.
6. **Ask via the host's structured-question tool** — 1–3 questions, 2–4 concrete clickable
   options each, recommended option first.
7. **Act immediately** on the answer, no re-confirmation.