Home

One Spec. 800 Apps. How Claude Code Enforces Consistency at Scale.

I just walked my dev manager through something that changes how we work across our entire portfolio. We manage over 800 mobile apps. Keeping implementations consistent across that many projects has always been a nightmare. Claude Code just fixed it.
The setup is dead simple. Open Claude Code from the project root. Run /init to generate your CLAUDE.md. Install your agents and slash commands into the root directory. That's your foundation.

Define the Spec Once

Here's the workflow I demoed. We picked an existing project where we'd already implemented AppOpen Ads by AdMob. I used my documentation agent — a custom subagent stored in .claude/agents/ — to describe exactly how we'd done it. Every detail. Every decision.
Then we talked about what we liked and what we didn't. What was working. What was clunky. What needed to change. The documentation agent captured all of it and produced a spec that described exactly what we wanted the implementation to look like going forward.
One conversation. One spec. Done.

Turn It Into an Auditor

Next step: we took that spec and created a new slash command called /app-open-auditor. If you haven't built one before, slash commands are just markdown files stored in .claude/commands/. You write what you want in natural language, and Claude executes it when you type the command.
Point /app-open-auditor at any project in our portfolio and it reviews the AppOpen Ads implementation against our defined standard. Finds the gaps. Implements the fixes.
That's 800+ apps that can now be audited and corrected against a single source of truth.
The key detail most people miss: after creating new agents or commands, you need to reload them in the local project so the new command is available. Show your team this step or they'll think something's broken.

Why This Matters

Before this, enforcing consistency meant documentation that nobody read, code reviews that caught some things but not others, and a lot of manual auditing. That doesn't scale. Not at 800 apps.
Claude Code's agents and skills flip the model. You define what good looks like once. You encode it as a command. Then any developer on your team can run that command against any project and get the same result.
This isn't just about AdMob. We can do this for any implementation pattern we want standardized. Analytics integration. Push notification setup. Authentication flows. Define the spec. Build the auditor. Run it everywhere.
One spec. 800 apps. Consistency without the meetings.

Get Started

If you want to try this workflow yourself:
  1. Install Claude Code and run /init in your project root
  2. Create custom subagents in .claude/agents/ for documentation and auditing
  3. Build slash commands in .claude/commands/ to codify your standards
  4. Check the best practices guide for tips on structuring your workflows
  5. Review the common workflows docs for more patterns you can adapt