Active development
In progress. Real code, not yet shown publicly anywhere else.
Design Bench · Holding AGI accountable
Agent skills & governance library
Reusable agent skills and change-control discipline for the Project-AI lab itself — a constitution, a deprecation policy, and claims-boundary docs versioned alongside the code they govern.
Active developmentinternal designation — agent_playbook
In more detail
Ed25519-signed evidence bundles and a 'no-silent-pass' validator back the change-control process. Its own README states it was integrated into Project-AI's main line 'per direct owner instruction' on a specific dated occasion — an explicit, first-party record of exactly when and why this stopped being a candidate and became load-bearing.
Concrete specifics
- ·Ed25519-signed evidence bundles for change approval
- ·A 'no-silent-pass' validator enforced in the pipeline, not just documented as policy
- ·README-documented owner-directed integration into Project-AI's main line, with a date
Go deeper