Vis Veritatis
v2-io is currently the umbrella for independent research and infrastructure under Joseph Wecker. Active work focuses on architectural and empirical methodology for aligned language-model-based agents.
- agentic-systems — formal architectural framework (Adaptation and Actuation Dynamics).
- hedge-embeddings — empirical research on language-model representation geometry.
- archema — declarative resource-oriented domain modeling (Ruby).
- autopax — production-quality agent-harness substrate (BLAKE3 hash-chained audit logs, self-authored system-prompt loading, session-continuation discipline).
- udon / libudon — markup notation designed for unambiguous reading and writing by language-model agents.
Repositories are public but have not yet been announced or circulated — by design, while the formal results consolidated. Orienting to academic engagement and peer review now.
At the moment, I haven't put anything up in the actual v2.io site. It is currently strictly a personal research umbrella.