agentleFS
SantiagoVqzSantiagoVqz/claude-skills/skills/support/setup-skills/SKILL.md

setup-skills

Skill2 starsChanged 12 days agoMITView on GitHub

Set up this repo for the skill set: tracker, labels, domain docs, worktree provisioning, and the Agent skills block in AGENTS.md. Run once per repo, again to repair, and with --upgrade to bring an earlier setup to the current rules.

  • Reads credentials

13 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use to-questionnaire, implement-spec and 8 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.
  • AWSSkill

    to-questionnaire

    Turn a decision you can't fully answer into a questionnaire for someone else to fill in.

    Also in 6 projects that use this

  • AWSSkill

    implement-spec

    Implement the result of /to-spec and /to-tickets in code.

    Also in 5 projects that use this

  • simo97Skill

    git-guardrails-claude-code

    Set up Claude Code hooks to block dangerous git commands (push, reset --hard, clean, branch -D, etc.) before they execute. Use when user wants to prevent destructive git operations, add git safety hooks, or block git push/reset in Claude Code.

    Also in 5 projects that use this

  • AnasX7Skill

    setup-pre-commit

    Set up Husky pre-commit hooks with lint-staged (Prettier), type checking, and tests in the current repo. Use when user wants to add pre-commit hooks, set up Husky, configure lint-staged, or add commit-time formatting/typechecking/testing.

    Also in 5 projects that use this

  • simo97Skill

    migrate-to-shoehorn

    Migrate test files from `as` type assertions to @total-typescript/shoehorn. Use when user mentions shoehorn, wants to replace `as` in tests, or needs partial test data.

    Also in 4 projects that use this

  • liufeng1976Skill

    setup-ts-deep-modules

    Wire dependency-cruiser into a TypeScript repo so each package is a deep module — implementation hidden in subfolders, reachable only through its entry-point files. User-invoked.

    Also in 3 projects that use this

  • +1 more from aws

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    Every other skill finds the tracker, the domain docs and the worktree rules through the ## Agent skills block in AGENTS.md.

    line 10
  2. Very specific:

    When the arguments hold --upgrade, follow UPGRADE.md in place of the steps below.

    line 12
  3. Distinctive:

    Read the repo's current state:

    line 16
  4. Very specific:

    When the repo needs provisioning: the filled CONFIG block of provision.sh, from PROVISIONING.md, and each change that its Compose rules …

    line 43
  5. Very specific:

    Labels: Use the Meaning in the table in tracker-github.md as the description.

    line 57
  6. Very specific:

    Worktrees: When the repo needs provisioning, install it with PROVISIONING.md.

    line 82
  7. Distinctive:

    Leave the changes uncommitted.

    line 86

“Common” is measured: kept by at least 6 in 10 of the 11 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

Setup skills

Every other skill finds the tracker, the domain docs and the worktree rules through the ## Agent skills block in AGENTS.md. This skill writes that block and the files it points at. A rerun repairs what is missing and keeps what is already correct.

When the arguments hold --upgrade, follow UPGRADE.md in place of the steps below.

1. Explore

Read the repo's current state:

  • git remote -v: the host of the repo.
  • gh auth status: gh can write to this repo.
  • AGENTS.md and CLAUDE.md at the root: which exist, whether one has an ## Agent skills block, and whether CLAUDE.md imports @AGENTS.md.
  • docs/agents/: each file, from an earlier run or from another skill set.
  • The glossary file: GLOSSARY.md, or CONTEXT.md when the repo keeps its terms there. With neither, it is GLOSSARY.md.
  • docs/adr/: each ADR that sets a rule for worktrees, provisioning or the tracker.
  • gh label list: labels that already exist.
  • .claude/skills/: each folder that copies a skill of this set under an old prefix, such as knot-build or sv-build, and each label description that names one.
  • What a new worktree lacks: gitignored .env files (git ls-files --others --ignored --exclude-standard), a lockfile, a database URL in an env file. When the repo has none of these, it needs no provisioning.
  • A .xcodeproj or a .xcworkspace: the repo is an Xcode repo.
  • scripts/provision.sh and .claude/hooks/provision-worktree.sh: files from an earlier run, or the repo's own script.

Then send one line: the tracker, and each conflict between the repo and the files that this skill writes.

Completion: you can name the tracker, the state of each file above, each label that is missing, and whether the repo needs provisioning. The one-line summary is in the chat.

2. Confirm

The tracker is GitHub when the remote is on GitHub. This version supports GitHub only. For any other tracker, stop and say that this version does not support it. Ask no question about the tracker, except when the repo has no remote: then ask "Create a private GitHub repo" or "Stop". On Create, run gh repo create <name> --private --source . --remote origin.

Show the user, in one message:

  • The tracker.
  • The ## Agent skills block from step 3, and the file that gets it.
  • The files that you will write or change, and the labels that you will create.
  • When the repo needs provisioning: the filled CONFIG block of provision.sh, from PROVISIONING.md, and each change that its Compose rules need.
  • Each existing file or label that you will keep as it is.

Send that message first. Then ask each choice in its own question tool call: each existing file or label that you could keep or replace. For each file in docs/agents/ that covers the same job as a file that you write, ask: "Merge", "Keep both" or "Delete". Put your recommendation first. For the old copies as a group, ask: "Delete" or "Keep". An agent picks a copy over the current skill, so recommend Delete. Recommend these:

  • A file that a commit deleted after an earlier run, such as the provisioning hook: "Leave deleted". The deletion is the user's choice. Give only the reason that the commit message states.
  • A file that a skill of the repo reads, such as a release skill: "Keep", also when no skill of this set reads it. Search .claude/skills/ for the file name. Then show the drafts, and ask for approval with the question tool. Give the options "Approve" and "Change". When the user picks Change, apply the edits and ask again.

Completion: the user picked Approve.

3. Write

Docs. Copy tracker-github.md to docs/agents/tracker.md and domain.md to docs/agents/domain.md. Write the glossary file from step 1 in place of <glossary>. When a file exists, keep the user's edits and add only what is missing.

Labels. Use the Meaning in the table in tracker-github.md as the description. For each missing label, run gh label create <name> --description "<meaning>". For a label that exists with another description, run gh label edit <name> --description "<meaning>". This keeps the label's color.

Block. The block goes in AGENTS.md, because agents other than Claude Code read that file. Make CLAUDE.md load it:

  • No AGENTS.md: create it with the block.
  • AGENTS.md exists: replace its ## Agent skills block, or append the block when there is none. Keep every other section.
  • No CLAUDE.md: create it with the line @AGENTS.md.
  • CLAUDE.md exists without @AGENTS.md: add the line at the top. When CLAUDE.md has an ## Agent skills block from an earlier set, move its content to AGENTS.md, so that one block remains.
## Agent skills

### Tracker

Ideas, specs, tasks and bugs are GitHub issues. To file an issue, use the `issue` skill. Before you read, link, label or close an issue, read `docs/agents/tracker.md`.

### Domain docs

Before you explore the code, read `<glossary>` and the ADRs in `docs/adr/` for the area. The rules are in `docs/agents/domain.md`.

### Worktrees

Code work runs in a linked worktree under `.claude/worktrees/`. The primary checkout stays on the trunk. Before code work starts on a branch, use the `worktree` skill.

Worktrees. When the repo needs provisioning, install it with PROVISIONING.md. Add .claude/worktrees/ to .gitignore in every repo.

Xcode. In an Xcode repo, follow XCODE.md.

Leave the changes uncommitted. The user commits them.

Completion: both docs files exist, gh label list shows every label from the table, AGENTS.md has exactly one ## Agent skills block, CLAUDE.md loads @AGENTS.md, and the provisioning check and Compose rules in PROVISIONING.md hold, or the repo needs no provisioning. An Xcode repo meets the completion of XCODE.md.

Report

Tracker · files written, changed or kept · labels created or updated · the file that holds the block · provisioning: deps, ports, database, or none · the files left to commit.

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.