Projects
Tools, platforms, and experiments.
Open-source projects ship from this lab with their own documentation sites. Platform engineering engagements are published as case studies within their confidentiality boundaries. Internal experiments graduate to projects when they earn it.
code-factorio/platform — AI platform engineering
Flagship engagement: the architecture of an internal AI platform — model gateway, agent runtime, evaluation harness, observability — and the boundaries an internal platform must respect. The case study publishes patterns and trade-offs; client identity and implementation details remain confidential.
Beskar — beskar.ventrion.dev
A curated, offline-first library for AI agent skills. Compose profiles per project, track changes with fingerprints, and deploy into any workspace. Rust CLI for Linux, macOS, and Windows.
Rohrpost — rohrpost.ventrion.dev
A git-native ticket system built for coding agents. Tickets travel with the code, every command emits JSON, and a single portable Rust binary replaces the server.
Experiments & internal research
Not everything ships on day one.
Smaller experiments — evaluation harnesses, agent runtime prototypes, workflow automation — live in private repositories until they are useful to anyone else. Findings get published in the writing section first; the code follows when it stabilizes. Nothing is announced here before it runs.