The Personal Agent Protocol Map (7 minute read)
Summary
Simon Taylor maps AI agent protocols across entry, identity, tools, and intent layers, arguing that banks and merchants need verified identities, limited permissions, and audit trails to safely authorize agent actions, while cautioning that access fees could slow adoption.
Similar Articles
Introducing Personal Agent Protocol (5 minute read)
Meta and Sierra, along with partners including Shopify, Stripe, and Walmart, announced the Personal Agent Protocol — an open standard defining how personal AI agents authenticate and interact with businesses via websites, APIs, or company agents, with a v0.1 spec planned soon.
@oviniciuslana: I couldn't understand the need to have a Personal Agent Protocol when one can work with an MCP with access control. By …
Bret Taylor announces the Personal Agent Protocol, an open standard developed with Meta, Sierra and partners like Shopify, Stripe and Walmart to define how personal agents interact with businesses, while a commenter questions whether MCP with access control already covers that need.
@changgaowei: Good summary. The customer decides what the agent may do, the company decides what it is allowed to do, and one file at…
Sierra, led by Bret Taylor, published the first draft of the Personal Agent Protocol (PAP, nicknamed Poppy) — an open standard for how personal AI agents shop and transact on behalf of users — and added 35 new partners including OpenAI, Visa, Mastercard, and PayPal, building on earlier support from Meta, Shopify, Stripe, and Walmart. The article breaks down PAP's core design, including .well-known/poppy.json service discovery, user-granted permissions, and agent self-identification, comparing it to MCP, A2A, and ANP.
@yatinbhalla42: Sierra and Meta proposed the Personal Agent Protocol, backed by Walmart, Shopify and Stripe. The key idea is the permis…
Sierra 和 Meta 提出了 Personal Agent Protocol（个人智能体协议），获得 Walmart、Shopify 和 Stripe 支持。其核心是权限模型：从访客身份起步，通过 OAuth 升级为已登录状态，且读写权限采用独立 scope——这意味着 Agent UX 已成为一个权限设计问题。
Personal-agent permissions need both requester identity and limits on returned data
The Open Instinct team open-sourced their personal-agent app and explains why tool allowlists alone fail the confused-deputy problem, proposing three checks: verifying requester identity from transport, authorizing the operation, and constraining returned fields before they reach lower-trust contexts.