agentleFS
petarzarkovpetarzarkov/dunx/.claude/skills/whats-next/SKILL.md

whats-next

Skill93 starsChanged 4 months agoMITView on GitHub

Write or refresh HANDOFF.md - a compact resume point holding completed objectives, live file paths, approaches already tried and rejected, and the exact next steps, placed against the roadmap in docs/ROADMAP.md. Use before /compact or /clear, at the end of a task block, when the context window crosses ~50%, or when asked "what's next", "where were we", "hand this off", or "resume".

16 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use handoff.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MicrosoftSkill

    handoff

    Compact the current conversation into a handoff document for another agent to pick up.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 14 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read the Open items section of docs/ROADMAP.md and the file under internal/notes/roadmap/ for the item in hand, when its row links one.

    line 20
  2. Distinctive:

    Add what only this session knows: what you actually changed, and what you tried that did not work.

    line 26
  3. Distinctive:

    If the session is thin on state - resumed, compacted, or you were handed a HANDOFF.md rather than doing the work - do not read packages …

    line 29
  4. Distinctive:

    Every line is a permanent tax on it.

    line 70
  5. Very specific:

    If an open item has been delivered, say so.

    line 79
  6. Distinctive:

    No conversation summary, no restatement of the plan, no preamble.

    line 81

“Common” is measured: kept by at least 6 in 10 of the 14 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/whats-next

Produce HANDOFF.md at the repo root: everything needed to resume this work in a fresh session and nothing else. It will be read cold, by a session with no memory of this one. Optimise for that reader, not for a record of what happened.

Gather

One batch, then stop:

git status --short && git log --oneline -8

Read the Open items section of docs/ROADMAP.md and the file under internal/notes/roadmap/ for the item in hand, when its row links one. That is the source of truth for what "next" means. Never invent a roadmap position from the conversation.

Add what only this session knows: what you actually changed, and what you tried that did not work.

If the session is thin on state - resumed, compacted, or you were handed a HANDOFF.md rather than doing the work - do not read packages into this thread to fill the gaps. Delegate one Explore subagent per unresolved question and ask each for a one-line verdict plus file:line. The exploration stays in their context window; only the verdict comes back.

Write

Overwrite HANDOFF.md with this shape. Omit a section that has no content - never write "N/A" or "none".

# HANDOFF - <YYYY-MM-DD>

**Roadmap:** <open item> - <what remains of it>
**Branch:** <branch> @ <short sha> · <clean | n files dirty>

## Done

- <objective> - <where it landed>

## Live files

- [path](path) - <what is half-finished in it>

## Rejected - do not retry

- <approach> - <what broke>

## Next

1. <exact command, or exact edit to an exact path>

## Open questions

- <question> - <what decides it>

Rules

  • Hard cap ~60 lines. This file is loaded at the start of the next session. Every line is a permanent tax on it.
  • Next steps are commands and edits, not intentions. "Add token() and inject() to packages/core/src/di/, resolving through a module-level currentInjector set around each new Klass()" - not "continue DI work".
  • Rejected is the highest-value section. It is the only content a fresh session cannot re-derive from the tree. Record the approach, the symptom, and the path it lives or lived in. Rejections already written down in docs/architecture/ belong there, not here - link, don't copy.
  • Paths as clickable relative markdown links.
  • If an open item has been delivered, say so. If docs/ROADMAP.md is stale, fix that doc - do not record the drift in HANDOFF.md.
  • No conversation summary, no restatement of the plan, no preamble.

After writing

State the path and the roadmap position in one line, then name the next move:

  • same subtask, context heavy → /compact, and say explicitly what to preserve
  • switching subtask → /clear, then continue from HANDOFF.md

HANDOFF.md is gitignored. It is session state, not a repo artifact - rewrite it freely.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.