Proposal: Agent-Native Department (ANDS) - Execution Semantics for Agent Systems from 陳信屹 on 2026-02-27 (public-s-agent-comm@w3.org from February 2026)

From: 陳信屹 <tyson@slashlife.ai>
Date: Fri, 27 Feb 2026 16:22:28 +0800
To: public-s-agent-comm@w3.org
Message-ID: <CAGRPLTPqNSbnvbhFkC5CeDpPW0i5_N_6bDtL=yy6s+XYRL97Pw@mail.gmail.com>

We have been working on a machine-executable execution model for
agent-based departments.

Key focus:

- task definition (machine-executable)
- constraint enforcement
- auditability (trace + replay)
- delegation model

Draft spec : https://github.com/slashlifeai/agent-native-department-spec

It is an execution semantics layer.

Would appreciate feedback on:

1. minimum task schema boundary
2. constraint enforcement model
3. audit / replay guarantees

Received on Friday, 27 February 2026 08:22:43 UTC