Nyxis: a Lean framework with formally proven authorization safety for AI-written backends
udmrzn · x · 2026-10-28
A Japanese developer open-sourced Nyxis (MIT, 290 GitHub stars), billed as the first backend framework that mathematically guarantees no data leaks. AI agents like Claude Code can write handlers freely; developers only manage a Cedar schema and authorization policies.
Written mostly in Lean, Nyxis carries a machine-checked proof that no sequence of HTTP requests can let an authenticated principal read, modify, delete data or grant permissions beyond their legitimate rights. Authorization flaws have topped OWASP's rankings since 2019; unlike Microsoft/GitHub's new PDD Kit, which requires per-program specs and human review, Nyxis fixes the theorem in advance so developers only manage policies.
Assumptions: trusted proof-external code and runtime; wrong policies still break protection. The author released it early due to the recent surge in security incidents.
More from coding & agent
- CodeRabbit Secure adds cross-repo Deep Scan, sandbox exploit verification and merge blocking — nevrekaraishwa2 · 2026-10-28
- Hyphenate raises $45M Series A after 86x revenue growth as agents run 90% of finance work — ahuja_priyank · 2026-10-28
- Open-source anymove turns one AI video into motion capture for any creature at $0.26 per clip — noahsolomon · 2026-10-28
- When to stop correcting an AI agent and restart the task instead — ZinchaDEV · 2026-10-28
- MASS paper: one model plays the whole agent team to bootstrap recursive self-improvement — idanbeck · 2026-10-28
- Dev builds AI-native macOS dock where an agent classifies activity and notifications in real time — trq212 · 2026-10-28