OpenAI debuts Agent Plugins, an open standard for bundling skills and MCP servers, and says Amazon, Cursor, Microsoft, and Vercel are on its steering committee
9to5MacZac Hall
Context & Ripple Effects
OpenAI had already moved from agent-building primitives to interoperability: its Responses API added remote MCP server support, and its Codex plugins program targeted repeatable workflows across more than 20 integrations. Agent Plugins extends that arc from individual integrations to a shared packaging format for skills and MCP servers.
The governance lineup also places OpenAI's initiative alongside Microsoft's Agent Control Specification, a separate open-source effort focused on limiting what agents may do. The two efforts address adjacent layers of agent deployment rather than the same problem.
First-order effects
Developers and MCP server operators gain an OpenAI-backed standard for bundling agent skills with server connections, replacing a purely one-off integration path.
Amazon, Cursor, Microsoft, and Vercel gain formal influence over the format through the steering committee, while OpenAI positions its earlier Codex plugin workflow model within a broader standard.
Second-order effects
Microsoft now participates in both a packaging initiative and an agent-control specification, increasing the incentive for developers to make skills portable while preserving granular controls.
Vercel, Cursor, and Amazon can shape compatibility requirements at the standard-setting stage, making implementation choices by agent-tool vendors more consequential than isolated API integrations.
Third-order effects
If major agent platforms converge on complementary standards for packaging and permissions, agent ecosystems may organize around interoperable components rather than platform-specific plugins.
Steering committees are becoming a key route by which infrastructure and application vendors influence agent distribution rules before those rules harden into developer defaults.
The trend: Agent platforms are moving from proprietary integrations toward shared standards that define both how capabilities are packaged and how they are governed.
i was very much against this it's a thin standard where most of the relevant stuff will be in client specific extensions anyway and it should not be coming from a company that does not even make an agent end users use
Oh we neeeeded this! 🙌🏾 I've been begging for a way to use my Skills and MCPs across the various harnesses I use. Amazon, Cursor, Microsoft, OpenAI, and Vercel got together and created a standard for Agent Plugins. Currently supported in Codex, Cursor, GitHub Copilot, Kiro,
We, the royal we, are still discovering what agents need to be effective. As these things are discovered, there is a history of them requiring new distribution and discovery mechanisms. The community had to invent mechanisms to install MCPs in coding agents. The community had
Devtools must be 1️⃣ open source and 2️⃣ universally extensible. AI coding agents are the most important devtools in the history of our industry. The Plugin standard lets anybody extend them uniformly. This is huge for the ecosystem. Build a Plugin → get exposure to the tidal
A much needed open standard for agent interoperability is here! 🎉 Agent Plugins gives Agent Skills and MCP servers a shared package format, so developers can build once and use the portable parts across compatible agent clients. The standard was developed by representatives
the ecosystems are getting locked sooner than ever before because you can build anything you want at an amazing speed so if execs align you can easily make the walled garden AI has made human relationships much more valuable in business now
Introducing Agent Plugins, an open standard for extending agents. Supports Agent Skills and MCP, with more to come. Built in collaboration with: @awsdevelopers, @code, @cursor_ai, @github, and @openaidevs. https://vercel.com/...
Build a plugin once and use it across compatible agent clients. Introducing Agent Plugins, an open standard developed with @awsdevelopers, @cursor_ai, @github, @code, and @vercel that packages Agent Skills and supports MCP server configurations in a shared format. [video]