domains / general / plugin-setup-wizard / setup

setup

Skill card

combined

All four copies are named 'setup' but they configure four unrelated systems: Power Automate CLI, the claude-ops plugin, an autoresearch experiment, and the Diverga research assistant. None is a reproduction of the others; each contains material steps and constraints the others lack, so crowning one would hide real alternatives.

Repositories

Other copies

Combined skill

# Setup

This skill is a generic setup wizard. Determine which target system the user is configuring, then follow the matching section. Do not mix steps from different systems.

## Power Automate CLI
1. Check Node.js with `node --version`. If missing or below 18, guide the user to install Node.js 18+ for their OS, then re-check.
2. Check Azure CLI with `az --version`. If missing, guide the user to install Azure CLI for their OS, then re-check.
3. Check Azure login with `az account show`. If not logged in, run `az login` and confirm with `az account show`.
4. Determine the Azure cloud with `az cloud show` and request a token for the matching Power Automate resource. Use `PA_CLOUD` and `PA_FLOW_RESOURCE` overrides for GCC, GCC High, or DoD tenants as documented.

## claude-ops Plugin
1. Run every Bash call with `run_in_background: true` and a short `description`.
2. Install missing CLIs and configure environment variables for each selected channel (Telegram, WhatsApp, Email, Slack, Notion, Linear, Sentry, Vercel).
3. Build the project registry and save user preferences to `$PREFS_PATH`.
4. Use `AskUserQuestion` for every decision; never auto-skip a channel the user selected.

## Autoresearch Experiment
1. If arguments are provided, pass them to `setup_experiment.py` with `--domain`, `--name`, `--target`, `--eval`, `--metric`, `--direction`, and optional `--evaluator` and `--scope`.
2. If no arguments are provided, collect domain, name, target file, eval command, metric, direction, evaluator, and scope one at a time, then run `setup_experiment.py`.
3. Report the experiment path, branch name, baseline metric, and suggest `/ar:run` or `/ar:loop`.

## Diverga Research Assistant
1. Check for an existing `diverga-config.json`; if present, ask whether to update.
2. Capture researcher discipline and experience level.
3. Capture statistical software and academic database access.
4. Configure Agent Teams and VS Arena mode, writing the selected settings to `diverga-config.json` and, if Agent Teams is enabled, setting `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` in `~/.claude/settings.json`.

## Conflicts
- microsoft/power-platform-skills::plugins/power-automate/skills/setup/SKILL.md requires Azure CLI and cloud-specific token configuration; the other copies do not.
- davepoon/buildwithclaude::plugins/claude-ops/skills/setup/SKILL.md requires `run_in_background: true` on every Bash call and messaging-channel configuration; the other copies do not.
- LeoYeAI/openclaw-master-skills::skills/autoresearch-agent/skills/setup/SKILL.md creates an autoresearch experiment via `setup_experiment.py`; the other copies do not.
- brycewang-stanford/Auto-Empirical-Research-Skills::skills/25-HosungYou-Diverga/skills/setup/SKILL.md captures a researcher profile and writes `diverga-config.json`; the other copies do not.