Built to be read, run, and questioned.
Each case study separates what works now from what is planned next.
Looper
A complete, pluggable Agent Runtime evolving from a minimal executable Core for real application development.
SearchX
Provider-aware WebSearch and safe WebFetch infrastructure, split into independently deployable Go services.
Token Atlas
A local-first, privacy-bounded observability dashboard for understanding AI tool activity across multiple coding clients.
Cache
A bilingual MkDocs knowledge base — AI/ML encyclopedia, Arch Linux notes, and classical annotation, kept current and published.
pi-agent-plugins
Pi Agent plugin monorepo — command-safety classifier (regex gate + model review, literal-string aware) and /btw, a based-on-context Q&A that answers without touching the main LLM context.
Public work joins the index when the evidence and limits are ready. Private systems stay outside.