agent-drift

​A neuro-symbolic runtime architecture for AI agents. Enforces zero trust privilege separation between probabilistic LLM reasoning and deterministic kernel execution. Prevents agent drift with real-time invariants and decision telemetry for auditable autonomy.

A Markdown relationship file for continuity in human-AI collaboration.

empirical research framework designed to analyze multi-agent Large Language Model (LLM) conversations as observable, continuous-state dynamical systems.

A durable Markdown orientation file for context recovery after memory resets and agent drift.

Free, local guardrail for AI coding agents — Claude Code, Cursor, Gemini CLI and more. Blocks out-of-scope writes, .env reads, rm -rf and force pushes. No API key, no account, no telemetry by default.