🤫husshhussh
🤫husshhusshOnePuppy
🤫IMPLEMENTATION GAP

DISTRIBUTED SYSTEMS · AUTHORIZATION · AGENTS

Can consent become a type system for agents?

Abstract. PCHP names identity, purpose, exact scope, validity, delivery, and receipt. The hard engineering question is whether those properties can become machine-checkable invariants across tool calls, instead of natural-language promises around them.

THE PROBLEM

How should scope, purpose, expiry, subject, and revocation compose so an agent cannot accidentally widen a grant while planning a multi-tool task?

HARD CONSTRAINTS

  1. 01A published registry and a token mint must share one vocabulary.
  2. 02Action tools must fail closed; read tools must not become covert write paths.
  3. 03Receipts must make a decision explainable without storing the private payload.
  4. 04Cross-device state cannot be honestly claimed before it has a durable, revocable model.

A FIRST CONTRIBUTION

Build a small, typed grant algebra and a property-based test suite: no composed plan may execute outside the intersection of its active grants.

BUILD IN THE OPEN

github.com/hushh-labs/hushh-research
hushh.ai/discord

PRIMARY SOURCE · hushh.ai/research/pchp-statusStatus report: real mechanisms and known gaps are both published.
QR code for https://www.hushh.ai/research/open-problems/consent-as-a-type-system
Read the researchhushh.ai/research/open-problems