agentleFS
chacosoldierchacosoldier/compabob/.claude/skills/setup/SKILL.md

setup

Skill34 starsChanged 21 days agoMITView on GitHub

One-time setup from inside Claude Code, no terminal needed: asks the setup questions in chat (assistant name, your name, role, language, kind of work, focus), then runs the kit's own setup.sh with the answers. Use for /setup, \"set up my assistant\", \"finish setup\", or when the session-start notice says setup is pending or unfinished. If setup already ran, it says so and names the files to edit instead of running again.

191 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use add-voice-transcription, add-discord 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.
  • juancruzmunozalbeloSkill

    add-voice-transcription

    Add voice message transcription to NanoClaw using OpenAI's Whisper API. Automatically transcribes WhatsApp voice notes so the agent can read and respond to them.

    Also in 5 projects that use this

  • MicrosoftSkill

    add-discord

    Add Discord bot channel integration via Chat SDK.

    Also in 11 projects that use this

  • MicrosoftSkill

    add-slack

    Add Slack channel integration via Chat SDK.

    Also in 11 projects that use this

  • juancruzmunozalbeloSkill

    add-telegram-swarm

    Add Agent Swarm (Teams) support to Telegram. Each subagent gets its own bot identity in the group. Requires Telegram channel to be set up first (use /add-telegram). Triggers on "agent swarm", "agent teams telegram", "telegram swarm", "bot pool".

    Also in 7 projects that use this

  • +8 more from microsoft

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 170 edited versions we read keep these.

  • 1

    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
  1. Very specific:

    All three exist and one of the four {{...}} tokens is left: an earlier run stopped early.

    line 14
  2. Very specific:

    Only _YOUR_NAME_HERE_ is left: ask only for the user's name, then run step 4 with just --user '<name>' --non-interactive.

    line 15
  3. Distinctive:

    If both or neither name a preset, use generalist and say so.

    line 28
  4. Very specific:

    Always pass --no-integrations: the integrations installer asks its own questions, which cannot be answered from here.

    line 38

“Common” is measured: kept by at least 6 in 10 of the 170 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

The same setup a terminal user gets from ./setup.sh, driven from the chat, for people who opened this folder in the Claude desktop app or VS Code. The script creates vault/, memory/ and config/user.config.yaml from the shipped seeds and writes the answers into them. It never overwrites a file that already exists.

Steps

  1. Check whether setup already ran, the same way setup.sh does. Look at vault/, memory/ and config/user.config.yaml, and Grep all three for {{ASSISTANT_NAME}}, {{USER_NAME}}, {{USER_ROLE}}, {{PRIMARY_LANGUAGE}} and _YOUR_NAME_HERE_.
    • Any of the three missing: a first setup. Go to step 2.
    • All three exist and one of the four {{...}} tokens is left: an earlier run stopped early. Say so in one line, then go to step 2 (the script fills only the blanks it finds).
    • Only _YOUR_NAME_HERE_ is left: ask only for the user's name, then run step 4 with just --user '<name>' --non-interactive.
    • None left: stop here. Say: "Setup already ran. I am <the assistant name in config/user.config.yaml>. To change your name, role or language, or what I know about your work, just tell me what to change and I will edit config/user.config.yaml and memory/ for you." Do not run the script.
  2. Ask about the user in one short message, numbered, defaults in brackets. They may answer "defaults" for any of them:
    1. What should I be called? [Aide]
    2. Your name? (no default; it goes into my memory)
    3. Your role or job title? [Knowledge Worker]
    4. Your main working language? [English]
    5. In a sentence or two, what do you work on? (may be left blank)
  3. Ask the kind of work with AskUserQuestion, one call with two single-select questions (the tool allows four options per question):
    • "Which describes your work best?": generalist (knowledge work broadly, the neutral default), consultant (client-facing delivery, proposals, stakeholder work), engineer (shipping software and owning technical decisions), "None of these, see the next question".
    • "Or one of these?": sales (pipeline, deals, accounts, quota), founder (running a business: strategy, fundraising, hiring, product), researcher (academia and R&D: literature, papers, experiments), "I already picked one above". If both or neither name a preset, use generalist and say so.
  4. Run the script. Build this command, each value in single quotes, any ' inside a value written as '\'':

    bash setup.sh --assistant '<a>' --user '<u>' --role '<r>' --language '<l>' --persona <id> --focus '<f>' --no-integrations --non-interactive

    Before running it, tell the user: "A permission box will appear asking to run bash setup.sh .... That is the kit's own setup script with your answers. Choose Yes (press Enter in a terminal, or click Allow in VS Code or the desktop app)." Then run it with the Bash tool, alone in that call.

    Always pass --no-integrations: the integrations installer asks its own questions, which cannot be answered from here. Integrations are optional; offer them in step 6.

  5. If the script stops with "Setup needs Python 3", nothing was created. On a Mac, offer to open Apple's installer for the free developer tools (they include Python): run xcode-select --install (one more permission box). A window opens; tell the user to click Install, wait until it says done (a few minutes), then tell you "done". Then run step 4 again with the same answers. On Linux, relay the script's install command word for word: it needs the user's password, so they run it themselves.
  6. Confirm from the files, not from the answers. Read config/user.config.yaml and report the assistant name, the user's name and the persona it holds. Read memory/MEMORY.md so you know the names in this session (the session-start hook only loads it in the next session). Then introduce yourself by your new name and offer the two steps that turn the kit into an assistant, in two lines, and do whichever the user picks (both are described in the constitution under "Looking after the kit"):

    • "Shall I connect your email and calendar? You click Connect once at claude.ai/customize/connectors, I guide you." After that, the user never pastes an email again.
    • "Shall I ask you about your projects, the people you work with and your recurring deadlines, and set up your knowledge base from that? About ten minutes."

    Then write today's date into .tmp/first-hour-nudge.done, so the next session does not repeat this offer.

    Mention once, briefly, that web and browser tools are optional: they need two free helper programs (Node.js and uv), and the user can ask "set up the web integrations" any time.

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.