Editorial illustration for GitHub repo maps AI tooling and system prompts around Claude Code
Claude Code Tooling Map Reveals Hidden Dev Secrets
GitHub repo maps AI tooling and system prompts around Claude Code
Ask an AI coding tool why it made a change, and you’ll likely get a vague rationale. The inner workings are typically hidden. That's changing.
Two public GitHub repositories are tearing down the curtain, publishing the internals of over a dozen popular assistants. They're not just documenting features—they're rethinking the entire concept of a coding copilot.
Asystem-prompts-and-models-of-ai-tools This repository is useful because it helps you understand the wider AI tooling landscape around Claude Code, not just Claude Code itself. The project collects exposed system prompts, tool definitions, and model-related details from a wide range of AI products, with the repository listing tools such as Claude Code, Cursor, Devin, Replit, Windsurf, Lovable, Perplexity, and others. That makes it especially valuable for people interested in prompt design, agent behavior, and comparing how different AI coding and productivity tools are actually structured behind the scenes, rather than only learning how to use one product in isolation.
gstack gstack is a strong example of how Claude Code can be used as a coordinated AI team rather than a single assistant. It reflects Garry Tan's Claude Code setup, with opinionated tools assigned to roles such as CEO, Designer, Engineering Manager, Release Manager, Doc Engineer, and quality assurance (QA), and the documentation shows these roles are structured through reusable skills and slash commands instead of ad hoc prompting. That makes it especially useful for anyone interested in role-based orchestration, more disciplined workflows, and a more team-like way of working with Claude Code.
get-shit-done If your goal is to work with Claude Code in a more structured way on larger projects, this repo is worth exploring.
The first repo, ‘system-prompts-and-models-of-ai-tools’, is a public autopsy. It gathers the exposed system prompts from products like Cursor, Devin, and Replit. This isn't a user guide.
It's a wiring diagram, showing the exact instructions baked into these tools before a user ever types a command. The second, Garry Tan's ‘gstack’, applies that knowledge radically. It rebuilds Claude Code as a full startup engineering org, with defined roles like CEO and Release Manager.
Each role gets specific tools. The shift is fundamental: from using a single, opaque assistant to engineering with a transparent, coordinated team. The machine's blueprint is out.
Now people are building factories.
Common Questions Answered
What makes this GitHub repository unique in mapping AI tooling around Claude Code?
The repository goes beyond typical code collections by curating system prompts, tool definitions, and model-related metadata from multiple AI products. It provides developers with a comprehensive view of the auxiliary components surrounding Claude Code, including tools like Cursor, Devin, Replit, and Perplexity.
Which AI tools are included in the GitHub repository's mapping?
The repository includes a diverse range of AI tools such as Claude Code, Cursor, Devin, Replit, Windsurf, Lovable, and Perplexity. By aggregating exposed system prompts and model details from these products, the project offers insights into the broader AI tooling landscape.
How comprehensive is the GitHub repository's coverage of AI tools and system prompts?
While the repository provides a substantial collection of prompts, templates, and tool definitions, it is not exhaustive. The project offers valuable insights and real examples of workflows and subagents, but the article acknowledges that the coverage is limited and does not specify the exact extent of the mapping.
Further Reading
- Leaked system prompts for 28+ AI coding tools hit 134K GitHub stars — Augment Code
- Claude Code Source Code Leak: 8 Hidden Features You Can Use ... — MindStudio
- Papers with Code - Latest NLP Research — Papers with Code
- Hugging Face Daily Papers — Hugging Face
- ArXiv CS.CL (Computation and Language) — ArXiv