Configure Rails for Conductor Parallel Agents

This skill sets up a Rails project to run correctly inside Conductor, the Mac app for parallel coding agents, so multiple agent workspaces get isolated ports, isolated Redis databases, and shared secrets without conflicting with each other. It creates conductor.json in the project root (if not already present) declaring a setup script (bin/conductor-setup) and a run script (script/server). It creates the executable bin/conductor-setup, which symlinks .env and the Rails config/master.key from the repo root - where secrets live, outside the per-workspace worktrees - and then runs bundle install and npm install. It creates the executable script/server, which sets PORT from $CONDUCTOR_PORT (defaulting to 3000), derives a separate Vite port by adding 1000, and, when $CONDUCTOR_WORKSPACE_NAME is set, hashes that workspace name with cksum to pick one of 16 isolated Redis database numbers exported as REDIS_URL, before executing bin/dev. For any of four existing Rails config files that already configure Redis - config/initializers/sidekiq.rb, config/cable.yml, config/environments/development.rb, and config/initializers/rack_attack.rb - it updates them to read ENV.fetch('REDIS_URL', ...) (or ENV['REDIS_URL']) with a sensible localhost fallback, so each workspace's isolated Redis DB is actually used. Implementation notes: it never overwrites conductor.json, bin/conductor-setup, or script/server if they already exist (it skips creation and informs the user instead), it only touches Redis-related configuration and skips files that don't exist or don't use Redis, and it creates the script/ directory if needed. Verification confirms all Conductor files exist and are executable, runs script/server to check it starts without errors, and checks that Rails configs properly reference ENV['REDIS_URL'] or the ENV.fetch fallback form.

Use this skill when configuring a Rails project to run correctly inside Conductor workspaces, needing isolated ports and Redis settings for parallel coding agents, or wanting the standard conductor.json/bin/conductor-setup/script/server scaffolding for a Rails repo. It is specific to Rails projects and Conductor's workspace model, not a general-purpose project scaffolding tool. Of the four Redis-configuring files it updates, config/cable.yml falls back to database 1 while the other three default to database 0.

Input is a Rails project that needs to run inside Conductor's parallel-workspace model. Output is a conductor.json manifest, an executable bin/conductor-setup that symlinks secrets and installs dependencies, an executable script/server that isolates ports and Redis per workspace, and updated Rails config files reading REDIS_URL from the environment with a fallback, skipped gracefully for any of the four config files that don't exist or don't use Redis, with the target Redis database chosen via the literal formula $((HASH % 16)).

Built for Conductor (the Mac app for parallel coding agents), targeting Rails projects using Bundler, npm, Sidekiq, Action Cable, Rails caching, and Rack::Attack, all reading Redis connection details from $REDIS_URL.

#!/bin/bash
set -e

### Symlink .env from repo root (where secrets live, outside worktrees)
[ -f "$CONDUCTOR_ROOT_PATH/.env" ] && ln -sf "$CONDUCTOR_ROOT_PATH/.env" .env

### Symlink Rails master key
[ -f "$CONDUCTOR_ROOT_PATH/config/master.key" ] && ln -sf "$CONDUCTOR_ROOT_PATH/config/master.key" config/master.key

### Install dependencies
bundle install
npm install

Rails developers using Conductor who need isolated ports, isolated Redis databases, and shared secrets scaffolded consistently across parallel agent workspaces.