Layered skill (index + on-demand references) for safe AI coding. 28 machine-enforced safety checks (SQLi, secrets, slopsquatting, reward hacking, doom loops). Runtime-agnostic: Hermes, Cursor, Codex, Cline, OpenClaw. MIT-0. KDS up to 83/100.
security owasp code-quality cursor ai-safety claude ai-code guardrails llm chatgpt vibe-coding coding-agent autonomous-coding loop-coding
-
Updated
Sep 1, 2026 - Python