agentleFS
princessingprincessing/claude-skills/skills/start/SKILL.md

start

Skill1 starsChanged 4 months agoMITView on GitHub

Orient and resume a chat for one topic — the bookend to /handoff. Resolves the topic to its git branch (or worktree), switches there, verifies you're on the RIGHT branch in a CLEAN tree (flags anything that looks like another chat's leftovers), fetches and fast-forwards from origin, and loads the latest handoff doc so you resume the destination instead of re-deriving the journey. Orient-only: never commits, never discards work. Use at the start of a session when the user says "/start <topic>", "start <topic>", "resume <topic>", "pick up <topic>", or opens a fresh tab to work a topic after a /clear.

130 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use setup-engine, team-audio and 10 more.

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.
  • CoralGameSkill

    setup-engine

    Configure the project's game engine and version. Pins the engine in CLAUDE.md, detects knowledge gaps, and populates engine reference docs via WebSearch when the version is beyond the LLM's training data.

    Also in 13 projects that use this

  • CoralGameSkill

    team-audio

    Orchestrate audio team: audio-director + sound-designer + technical-artist + gameplay-programmer for full audio pipeline from direction to implementation.

    Also in 13 projects that use this

  • lilangMaxSkill

    create-stories

    Break a single epic into implementable story files. Reads the epic, its GDD, governing ADRs, and control manifest. Each story embeds its GDD requirement TR-ID, ADR guidance, acceptance criteria, story type, and test evidence path. Run after /create-epics for each epic.

    Also in 11 projects that use this

  • lilangMaxSkill

    story-done

    End-of-story completion review. Reads the story file, verifies each acceptance criterion against the implementation, checks for GDD/ADR deviations, prompts code review, updates story status to Complete, and surfaces the next ready story from the sprint.

    Also in 11 projects that use this

  • +5 more from CoralGame
  • +3 more from lilangMax

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 117 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    gated: if the topic has no branch/worktree yet, offer to create one — shown in full, run only on an explicit yes.

    line 19
  2. Very specific:

    Read the newest match; confirm its title/DESTINATION DOC line names the expected topic before trusting it (guards against a …

    line 85
  3. Distinctive:

    Treat a handoff as a map: read the docs it names (and the docs THOSE name) before claiming the picture is loaded.

    line 89
  4. Distinctive:

    Don't work in the wrong folder, and don't edit two topics from one chat.

    line 106
  5. Very specific:

    Don't stash/reset/checkout/clean/-A anything as part of orientation — surface it and ask.

    line 107
  6. Distinctive:

    Don't pull when dirty; don't auto-merge/rebase a divergence.

    line 108
  7. Very specific:

    Don't commit, push, or change code — /start is read + fast-forward only (plus the one gated create-worktree offer).

    line 109

“Common” is measured: kept by at least 6 in 10 of the 117 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

Start / resume a topic chat — orient before you touch anything

The opening bookend of the session lifecycle (the mirror of /handoff, the closing one). If you run several topics in parallel — each on its own branch, often its own git worktree — a fresh chat has zero context and the #1 failure mode is working in the wrong folder/branch or on top of another chat's half-finished edits. This skill removes that: it lands you on the right branch, confirms the tree is clean (or surfaces exactly what's dirty and why), syncs with origin, and hands you the resume summary from the latest handoff doc.

/start is orient-FIRST. It may git fetch and fast-forward pull, and it READS docs — but it never commits, never pushes, never discards another chat's work. The one write it may do is gated: if the topic has no branch/worktree yet, offer to create one — shown in full, run only on an explicit yes.

Step 1 — Resolve the branch / worktree

Figure out where this topic lives:

git worktree list          # if you use one worktree per topic
git branch --list          # otherwise, the topic's branch
  • Topic maps to a worktree → that folder is where you work; all later git ops use git -C <worktree> … and edits use absolute paths under it.
  • Topic maps to just a branch → git switch <branch> (only if the current tree is clean; if dirty, STOP and report — don't switch over uncommitted work).
  • No topic given → list the worktrees/branches and ask which one; don't guess.
  • No branch/worktree yet (new topic) → offer to create it (gated — show the command, run on yes):
    # branch only:
    git switch -c <topic>
    # or a dedicated worktree off your integration branch:
    git worktree add -b <topic> ../<topic> <base-branch>

Step 2 — Clean-tree check (the contamination gate — the point of this skill)

git -C <worktree> status --porcelain

Classify — do not auto-clean anything:

  • Empty → clean. Say so and continue.
  • Dirty with THIS topic's files → likely legit in-progress work the last handoff mentions. Cross-check against the handoff (Step 4); if it lists these as "uncommitted", report as resume-in-progress and leave it.
  • Dirty with files that don't fit this topic → contamination flag. Surface the exact files, say they look out of place, and ask whether to leave, move, or (only on explicit say-so) discard. NEVER git stash/reset/checkout/clean without an explicit instruction — another chat may own that work.

Step 3 — Sync with origin (safe, fast-forward only)

git -C <worktree> fetch origin
git -C <worktree> status -sb        # read ahead/behind vs origin/<branch>
  • No upstream (fresh branch) → report it and skip the pull (a bare pull --ff-only errors "no tracking information"). /commitpush will set the upstream when you first push.
  • Behind & clean → git -C <worktree> pull --ff-only. Report the new tip.
  • Behind & dirty → do NOT pull (could conflict); report "behind by N, pull after committing."
  • Ahead → note "N local commits not pushed" (normal).
  • Diverged → report it; don't auto-merge/rebase — let the user decide.

Step 4 — Load the resume context (handoff → destination doc)

Find the newest handoff doc for this topic (see the /handoff skill — default folder handoffs/, overridable with HANDOFF_DIR):

HO="${HANDOFF_DIR:-$(git rev-parse --show-toplevel)/handoffs}"
ls -t "$HO" | grep -iE '<topic>'      # newest first
  • Read the newest match; confirm its title/DESTINATION DOC line names the expected topic before trusting it (guards against a newest-but-mismatched file). Follow its DESTINATION pointer and read that doc too.
  • No handoff found? Say so and fall back to the topic's README/plan + recent git log.
  • Treat a handoff as a map: read the docs it names (and the docs THOSE name) before claiming the picture is loaded.

Step 5 — Report the orientation (concise, actionable)

  • Folder + branch: "Working in <worktree> on <branch>."
  • Tree state: clean ✅ / resume-in-progress (list files) / ⚠️ looks-contaminated (list + what you asked).
  • Sync: ahead/behind vs origin; whether you fast-forwarded.
  • Where we are / what's left: 3–6 bullets from the handoff + destination doc — current state, what's done, and the next concrete task(s).

Then stop and let the user direct the work. Do not start editing as part of /start — orientation only.

Don'ts

  • Don't work in the wrong folder, and don't edit two topics from one chat.
  • Don't stash/reset/checkout/clean/-A anything as part of orientation — surface it and ask.
  • Don't pull when dirty; don't auto-merge/rebase a divergence.
  • Don't commit, push, or change code — /start is read + fast-forward only (plus the one gated create-worktree offer).

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.