Studio inventory / Software & experiments
A small collection of tools and systems. Public source where available; explicit limits where it isn’t. None of these entries is a promise of a finished product.
A local-first coding harness. Typed tools, bounded agent loops, and a verification gate before the work is called finished.
An experimental Go harness for llama.cpp, not a general claim of autonomous coding reliability.
A show engine that treats sound and picture as things to assemble carefully: frozen inputs, explicit timelines, and checked outputs.
CPU montage assembly is implemented. GPU rendering, music generation, voice cloning, and native lipsync are roadmap items.
One model, one code-golf problem, one independently judged program. A small test bench for inspecting agent results rather than trusting the transcript.
Early Python-only prototype for static-output problems. No public release or benchmark leaderboard is offered here.
Graphic studies by the studio. These are visual identities, not software screenshots or measured results.
A source-backed knowledge workflow inspired by Andrej Karpathy’s LLM Wiki pattern: keep original sources, connect working notes, and carry citations into the synthesis.
No public release. The working collection is private; its notes, operational records, and source files are not published here.
The evidence principle ↗Reading the labels
Public source means there is a repository to inspect, not a supported hosted service. Early alpha means a narrow foundation exists and significant capabilities are still missing. Local prototype means local implementation, without a public release.
Descriptions are based on inspected project documentation. This site does not independently certify every capability or publish benchmark claims.
Historical notes & corrections ↗