process-setup

/process-setup — one-time per-repo configuration

You are running the /process-setup skill for the claude-slack-bridge full-process plugin. This is a one-time-per-repo configuration flow. It does NOT start a feature — it only writes the template, optional helper skill, and .gitignore entries that /process will need later.

You run locally inside Claude Code (not via the Slack daemon). All clarifications go through AskUserQuestion. Do not call mcp__claude-slack-bridge__ask_on_slack from this skill — Slack is only the runtime channel for /process itself, not for setup.

The plugin and the daemon's workflow engine are version-locked. The template you write below has version: 1; the daemon checks this on every step spawn and refuses to advance if the version is unsupported. Do not invent a different version.

Step 1 — verify mcp__claude-slack-bridge is installed in the repo (delegated)

Delegate this check to the verify-bridge skill in a separate context. Spawn it via the Agent tool with run_in_background: true so the verification chatter does not pollute this orchestrator's context window:

Agent({
  description: "Verify Slack bridge MCP installed",
  subagent_type: "general-purpose",
  prompt: "Read plugin/skills/verify-bridge/SKILL.md (resolve the plugin root via ${CLAUDE_PLUGIN_ROOT} if set, otherwise search upward from cwd until you find plugin.json). Follow its instructions exactly: read cwd/.mcp.json and confirm a `claude-slack-bridge` entry exists under `mcpServers`. On success, return the literal string 'verify-bridge: ok'. On failure, return the exact fix-it message from the skill and report non-zero status. Do not write any files, do not modify .mcp.json, do not check container runtime health — only verify the declaration.",
  run_in_background: true
})

When the subagent's completion notification arrives, branch on the result:

• ok — continue to Step 2.
• failure — print the subagent's returned fix-it message verbatim to the user and exit non-zero. Do not write any files, do not proceed.