Aakash Gupta and The Information point to an accidental Claude Code leak exposing unreleased feature flags and a background agent, which rivals may use as a roadmap and which raises supply chain style concerns about what ships inside tooling.
Competitors are already reacting after Anthropic’s leak exposed upcoming features for its Claude Code agent.
Anthropic Mythos code leaks. How bad is it?
We are pushing code to production faster and faster.
And our AI Agents are pushing code autonomously everywhere, to open databases, to unlocked repositories.
Anthropic built an internal system called "Undercover Mode" specifically to prevent their AI from leaking internal information into external contexts.
Then they shipped the entire 512,000-line codebase in a .map file because someone forgot to update .npmignore.
The leaked
Anthropic’s accidental code leak is offering competitors a roadmap for its next push in AI agents.
Some rivals say they may speed up their own launches as a result.
A source map file shipped inside the Claude Code npm package this quarter. 512,000 lines of TypeScript. 44 unreleased feature flags.
A persistent background agent called KAIROS that nobody was supposed to know about.
Anthropic s accidental code leak is offering competitors a roadmap for its next push in AI agents.
This finding is one of many signals tracked across Indiehacking. The live feed updates every few hours with new authority voices, debates, and emerging ideas.
← Back to Indiehacking