Why pyhall

Governance that proves itself.

Every AI worker decision should be auditable, policy-driven, and verifiable. Not eventually. Every time.

Govern

Define who can run what worker, at what capability level, under which policy. No ambiguity — workers operate inside declared boundaries or not at all.

Worker Class docs →

Attest

Every worker carries a signed capability card. Every decision produces a verifiable proof. You know exactly what ran, what it was authorized to do, and when.

Attestation docs →

Route

Match work to workers by capability, policy tier, and availability. Mismatches are rejected at the gate — not silently degraded at runtime.

Routing docs →

Observe

Every routing decision is logged with proof. Query history, audit decisions, trace incidents. Live worker health in one monitor view.

Live monitor →

How it works

Four steps from zero to governed AI workers.

01
PUBLISH

Define a worker with a capability card.

hall publish worker.wcp.yaml
02
ENROLL

Register with the pyhall registry. Claim your namespace.

hall enroll --namespace x.you
03
ROUTE

Send work through the Hall. Policy enforces who runs it.

hall route --task "analyze data"
04
OBSERVE

Every routing decision is logged with proof.

hall monitor --worker x.you.analyst