Claude Code Source Leak Reveals Unreleased Features and Internal Roadmap

TLDR

Anthropic's Claude Code client application source code was accidentally leaked, revealing extensive details about the product's internal roadmap. The codebase, internally codenamed 'Tengu', contains over 60 feature flags—switches that enable unreleased functionality for specific users—indicating capabilities that are built but not yet publicly available.

Analysis suggests Anthropic is positioning Claude Code as a cloud-based, vertically-integrated version of OpenClaw, the AI agent framework that gained prominence earlier in 2026. While some discovered features appear nearly ready to ship, others remain experimental. The leak involves client application code rather than model weights or training data, but provides unprecedented visibility into Claude Code's development direction.

Key Takeaways

• Anthropic's Claude Code source code was accidentally exposed, revealing 60+ feature flags, internal codename 'Tengu', and plans positioning it as a cloud-based OpenClaw alternative