Audit receipts
Tamper-evident traces covering who delegated, what was allowed, what happened, and whether scope held.
The cross-boundary identity specification for AI agents: identity, authentication, authorization, and audit across systems.
Agent Residency.
The briefing is the artifact. The page makes it move.
A light preview, not a second report. Drag the surfaces, then open the PDF for the complete specification.
Per-agent identifiers, principal binding, legal-entity anchor, and proof that an agent is not a borrowed user.
Agent login without human credential inheritance. The agent enters as itself, not as a person wearing a mask.
Machine-readable mandates: limits, expiry, attenuation, revocation, approval states, and refusal boundaries.
Tamper-evident traces covering who delegated, what was allowed, what happened, and whether scope held.
Sub-agent provenance and monotonic authority: every hop carries less or equal authority than the one before it.
Revocation, review, standards alignment, and institutional controls around deployed agent authority.
One interaction, not another explainer: run an action and watch authorization fail when the mandate is missing.
Pay new supplier EUR 200,000
Awaiting audit seal.
That was the diagnostic. The briefing maps the full coverage and the pilot path.
V will map identity, authentication, authorization, and audit around one regulated workflow — including the live revocation drill the current AI stack cannot run. The briefing is the shared starting point.
The specification and pilot framework are ready for review. Engagement now has concrete paths.
Review the specification and pilot framework for institutional fit, liability boundaries, and public-sector responsibilities.
Review the mandate, revocation, audit, and liability model.
Evaluate verifier-SDK integration, mandate checks, transaction limits, and account-control boundaries.
Open a pilot conversation, or download the briefing.