Control your Claude Code sessions from your phone — live status cards, a per-session chat view, a real terminal, and push notifications with approve/deny, all served over Tailscale.
Security tool that analyzes GitHub repositories for prompt injection vulnerabilities and attack hotspots.
RL-powered research paper triage over iMessage using a contextual bandit and a local Ollama LLM — zero API costs.
Model-agnostic auditor that quantifies how much of an ML model's benchmark score comes from data leakage, then recomputes an honest score — without ever running the model.