agentleFS
samsam007bsamsam007b/claude-code-standards-kit/skills/welcome/SKILL.md

welcome

Skill2 starsChanged 8 months agoMITView on GitHub

First-run onboarding for a beginner. Assesses level and needs in 6 questions, explains each tool before installing it, installs only after a yes, then writes the profile.

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 5

    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. Distinctive:

    If the person asks a side question, answer it, then ask the pending question once, plainly.

    line 20
  2. Distinctive:

    If a read needs permission, say "I ask your permission before reading", never "blocked".

    line 21
  3. Distinctive:

    Never guess the level: ask, then confirm in one sentence ("So you are at level Practice, correct?").

    line 22
  4. Very specific:

    Read ~/.claude/kit/profile.json.

    line 26
  5. Distinctive:

    Do you work with mail, calendar, documents or other apps you would like Claude to reach?

    line 34
  6. Distinctive:

    Do you need Claude to browse or test web pages for you?

    line 35
  7. Very specific:

    Write ~/.claude/kit/profile.json with language, level, needs, tools_installed, mode: "learner", date.

    line 62

“Common” is measured: kept by at least 6 in 10 of the 7 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

/welcome

Language: if the person writes in French, or language is fr in ~/.claude/kit/profile.json, follow SKILL.fr.md in this folder instead.

Didactic path for someone setting up Claude Code alone (or after a short training). Skipped by /pro-samuel.

Principles

  • Simple words. Define any technical term the first time, or point to /explain <term>.
  • Nothing is installed without a clear "yes". Before each tool: what it is, why it helps this person, what it changes on the machine, how to remove it.
  • One question at a time. Never more than 6 in total.
  • If the person asks a side question, answer it, then ask the pending question once, plainly. Never apologise for repeating it or promise it is the last time.
  • If a read needs permission, say "I ask your permission before reading", never "blocked". If the profile cannot be read, answer in the language of the person's message.
  • Never guess the level: ask, then confirm in one sentence ("So you are at level Practice, correct?").

Step 1: Detect the setup (no questions yet)

Read ~/.claude/kit/profile.json. If it exists and --redo is absent, greet by level and ask what to do today instead of restarting. Otherwise detect: OS, shell, and presence of Homebrew or winget, Git, Node, Python. Report in plain words what is already there.

Step 2: Six questions

  1. What do you want to do with Claude Code? (write, code, automate, analyse, other)
  2. Have you used a terminal before? (never / a little / often)
  3. Have you used Claude Code before? (never / a few times / daily)
  4. Do you work with mail, calendar, documents or other apps you would like Claude to reach?
  5. Do you need Claude to browse or test web pages for you?
  6. How do you prefer to learn: short explanations, or detail with examples?

Step 3: Level

LevelSignalEffect
Discoverynever used a terminal or Claude CodeExplain everything, propose few tools, extra safeguards
Practicesome terminal or Claude Code useShort explanations, tools proposed from needs
Autonomousdaily useTerse, option lists, advanced tier offered

Step 4: Propose tools from needs, one by one

NeedToolPlain explanation
Browse or test pagesPlaywrightLets Claude drive a Chrome window like a person would
Mail, calendar, docs, many appsComposioOne account that connects Claude to hundreds of apps without wiring each one
Any person, any levelComposio, offered to everyoneOptional: the benefit is explained, a free account is needed, refusal costs nothing
Reusable routinesSkillsA written recipe Claude can follow on demand
Safety netGuardrails (2 hooks)Small checks that stop risky commands before they run

For each: explain, ask "install it?", install on yes, verify it works, record in profile.

Guardrails: on yes, run node ~/.claude/kit/hooks/register-hooks.mjs (backs up settings.json, merges, undo with --remove). Say what they block: destructive commands (guard-destructive) and real-looking secrets written to files (guard-secrets). Pro mode registers them in the single confirmed pass.

Step 5: Write the profile

Write ~/.claude/kit/profile.json with language, level, needs, tools_installed, mode: "learner", date. Local file, never sent anywhere.

Step 6: Close

One short summary: what is installed, what each does, how to ask for help (/explain <term>), how to change level (/welcome --redo). Offer the advanced tier only to Autonomous, and say how: keep the kit folder you downloaded (the starter copies only 3 skills), then run bash global/install-global.sh --dry-run to preview and without --dry-run to install (backs up, never overwrites). Everyone else: mention once that /welcome --redo later unlocks it.

Rules

  • Windows: tier 1 runs natively. The advanced tier needs WSL2; explain it and let the person decide.
  • Language follows language in the profile; the French edition mirrors this file.
  • Any external send or account creation needs a confirmation in the same turn.

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.