AURA-Lab
The builder's workbench, and the reference implementation of Execution Governance Dynamics.
Where it stands: Working prototype
2,025 source files across 15 commits. Last worked 20 July 2026.
AURA-Lab is where the governance ideas the rest of this ecosystem depends on are actually implemented. Its sentinel_core crate provides four primitives: context binding, which cryptographically ties a granted capability to the state of the environment it was granted in; continuous trajectory risk scoring; rate shaping through verifiable delay functions, so an agent cannot simply act faster than oversight; and a graduated intervention hierarchy instead of a single kill switch.
This matters commercially because it is the same discipline the shipping products are held to. The Sentinel is not a safety layer bolted onto AURA-Lab after the fact — AURA-Lab is the reference implementation of it.
Next step
Harden the crate's public API and publish the governance primitives as a standalone dependency the other projects consume.
Interested in this one specifically? Get in touch.
The rest of the directory
- The core system
- Making things, real and rendered
- Moving data, and proving truth
- Human discipline, and open experiments
- Research and outreach