Claude Code Team Commander — Security skill for Claude Code

Claude Code Team Blueprint

**A complete playbook for building a shared Claude Code configuration for your company.**

The Problem

When you give your team Claude Code, everyone sets it up differently. Different plugins, different prompts, no shared standards, API keys scattered everywhere, and no way to control what tools are available. New hires spend hours figuring out the setup instead of doing work.

What We Built

At [OIT, LLC](https://oit.co), we run a VoIP business with a Salesforce org, internal knowledge base, VoIP platform APIs, Microsoft 365, and a growing team. We needed Claude Code to work the same way for everyone — with our SOPs, our quality standards, and our integrations baked in — while keeping secrets secure and giving admins control over what tools the team can use.

We built a single GitHub repo that does all of this:

• 5-minute onboarding — one script installs everything and prompts for API keys
• AI personas that know our workflows — case quality scoring, Salesforce automation, documentation standards, support ticket handling, full-stack development
• 18 slash commands that trigger specific workflows — /stan-review 222448 scores a case against 8 quality dimensions, /flow-review Lead_Assignment audits a Salesforce flow for governor limits
• Admin-controlled catalog of plugins and MCP servers — nothing enters the toolkit without a PR approval
• Auto-discovery — any team repo can propose a new MCP server by adding a .claude-catalog-entry.json file; a weekly GitHub Action scans for these and creates a PR for admin review
• Announcement system — admins post one-time messages that display the next time a team member syncs
• Security by default — API keys never committed, secret scanning with push protection, branch protection requiring PR approval, TruffleHog CI on every push

How It Works

1. Team member clones the config repo
2. Runs setup.sh in their project directory
3. Script copies personas/c