cytostack/openwolf — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me an open source TypeScript command line tool called OpenWolf that acts like a second brain for Claude Code. I want someone to install it globally, run openwolf init inside any project, and then keep using Claude normally while OpenWolf quietly helps in the background.

It should create a .wolf folder with a project map, learned preferences, a memory log, bug fix history, a token ledger, config, identity, and Claude instructions. The big goal is to save tokens by telling Claude what files contain before it reads them, warning when it rereads the same file, remembering corrections, and updating the project map after code changes.

Include the basic commands like init, status, scan, dashboard, daemon start and stop, bug search, restore, update, and a design review command that captures screenshots of a running app for Claude to inspect. Keep the setup simple, Node 20 plus, with sensible defaults and no workflow changes for the user. Look up current Claude Code hook docs if you need to.

Want more depth? Deep Reverse