coleam00

coleam00 / helpline

Public

Demonstration codebase for building the AI Layer (CLAUDE.md hierarchy, hooks, skills, LSP, MCP, plugin) — companion repo for the 'How Claude Code works in large codebases' video.

11
5
100% credibility
Found May 21, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Helpline is a demonstration project showing how to build an 'AI Layer' - a configuration and tooling harness that makes AI coding assistants productive in large codebases. It includes a realistic B2B helpdesk platform (with ticket management, user accounts, billing, and notifications) alongside complete implementations of hooks, skills, structured search tools, and self-improving documentation. The project serves as both a working application and a teaching tool, with two learning paths: install the portable plugin directly, or study the patterns to build your own AI Layer adapted to your project.

How It Works

1
💡 You want an AI coding assistant that actually understands your project

You've tried asking an AI to help with your code, but it gets lost in large projects and doesn't know your team's conventions.

2
📦 You discover a complete example that shows how it's done

This project is a real helpdesk application with all the configuration and tooling that makes an AI assistant productive in a complex codebase.

3
🎫 You explore the helpdesk platform itself

The project includes a working helpdesk with ticket management, user accounts, billing, and email notifications - everything a small support team needs.

4
You choose how to learn from this project
🚀
Install the plugin into your project

Take the portable pieces - hooks, search tools, and skills - and add them to your own codebase with one command.

📖
Study the patterns and rebuild

Read the documentation and rebuild the AI Layer from scratch, adapted to your own project structure.

5
🧩 Your AI assistant gains superpowers

The AI now knows your project structure, follows your team's conventions, and can search your code structurally instead of just guessing.

6
🔄 The AI Layer improves itself over time

After each session, the AI reflects on what changed and proposes updates to keep the documentation fresh and accurate.

Your AI coding assistant finally works the way you imagined

You have an AI partner that understands your codebase, respects your team's patterns, and gets smarter with every session.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is helpline?

Helpline is a demonstration codebase built to teach developers how to give Claude Code the context and tooling it needs to work effectively in large, multi-service projects. The repo implements a B2B helpdesk platform with five services (API gateway, authentication, billing, notifications, and search) sharing common packages. But the actual product here is the AI Layer itself: a hierarchy of CLAUDE.md files, hooks that orient the agent at session start and reflect on sessions to propose improvements, path-scoped skills for domain-specific tasks, an AST-based search server, and a portable plugin for one-command installation into any project. Everything is built in Python and validated end-to-end with a test suite.

Why is it gaining traction?

Anthropic's article on large codebases describes what an AI Layer should look like, but stays high-level. Helpline is the worked example that shows every component actually built and running. The self-improving Stop hook is the standout feature: after each session, it reflects on what changed and proposes CLAUDE.md updates so your instructions never drift from reality. The AST-based search server moves past text matching to structural queries about definitions and references. The plugin bundles the reusable pieces so you can install the whole baseline into your own repo in one command.

Who should use this?

Developers running Claude Code on mid-to-large Python codebases who want to stop fighting the agent and start shipping. Teams with multiple services or shared packages will get the most value from the CLAUDE.md hierarchy and path-scoped skills. If you've tried prompting Claude to follow conventions and it keeps forgetting, this repo shows how to make that stick. Not for beginners or single-file scripts.

Verdict

At 11 stars, this is early-stage and the credibility score sits at 1.0%, so treat it as a learning resource rather than a production-ready framework. The documentation is thorough and the validation suite proves every piece works, but the helpdesk itself is deliberately minimal. The real value is the pattern library in the tooling directory. Clone it, read AI-LAYER.md, and have Claude build you a comparable layer for your own codebase.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.