Multi-agent systems · 2026
Governed Agent Engine
A governed multi-agent engine: Termux-installable, stdlib plus PyYAML, with spec binding and audit.
The Governed Agent Engine is deliberately small: standard library plus PyYAML, installable on Termux, with a $0 free-model cognition fallback so cost is not the reason a governed run does not happen.
Governance is not a document next to the system. The spec binds the run, and the audit trail shows where the run followed it and where it did not.
It is the engine behind the ForgingStack position: if an agent cannot be audited, it cannot be trusted with work that matters.
More work
Agent reliability · Open source
EvidenceForge
An open-source framework for building auditable, evidence-citing AI agents.
API tooling · Developer product
mimicplus
Record a HAR, get a versioned spec, a typed Python client, OpenAPI, and a breaking-change check.
Audit infrastructure
Verity
An append-only evidence ledger for agent runs. Claims must cite evidence.