Origin - Your AI agents build fast. Origin keeps them in check.
by•
AI agents write most of your code now. But nobody knows which lines they wrote, what prompted them, or what it cost.
Origin fixes that.
Open source CLI tracks every AI session - Claude Code, Cursor, Copilot, Gemini. Line-by-line attribution in your git repo.
getorigin.io (http://getorigin.io/) adds governance for CTOs: session replay, cost tracking, policy enforcement, PR compliance checks.
Built by a DevOps guy with no coding background. Just an idea and Claude Code.


Replies
Curious about the cost tracking across different AI tools. We're juggling Claude Code, Cursor, and Copilot on different projects and have little to none visibility into what we're actually spending. Does Origin aggregate costs across all the tools or track them separately?
@f_begiello thanks for your interest. Yes, Origin aggregates costs across all agents in one dashboard. Claude Code, Cursor, Codex, Gemini - each session tracked separately but you see the total in one place: cost per agent, per developer, per repo, per day.
Copilot is trickier since Microsoft doesn't expose token/cost data via hooks yet. For Claude Code and Cursor you get full cost breakdown down to individual sessions and prompts.
The product idea is good but it is not reliable yet. I have tested it on 3 of my projects.
project 1: manual + qwen code. showing all changes are made by human
project 2: manual + claude haiku(via Zed github copliot integration). showing all changes are made by human.
project 3: gemini cli(entire project). showing all lines written by human.
@dshaw0004 thanks for the comment. The Origin will only show AI generated code committed after you installed Origin CLI. There is no way for now to backtrack existing code. If you have ideas how to do it - we can discuss.