Multi-agent decision lab with visible votes, uncertainty, provenance, receipts, and human execution gates

Train AI agents on tools, validate real work, and publish portable mastery evidence

Static, offline supply-chain airlock for repositories, packages, agent bundles, and CI workflows.

Local-first evidence hashing and portable build receipts for public work

Evidence-backed claim production with credentialed workers and independent verification factories.

Compile quorum-approved, safety-reviewed agent proposals into bounded engineering work packets