Amazon, Cursor, Microsoft, OpenAI, Vercel unite on an AI agent plugin standard

Amazon, Cursor, Microsoft, OpenAI, and Vercel jointly released Agent Plugins, an open, vendor-neutral standard defining a single package format for AI agent extensions. Version 1.0.0 uses a plugin.json manifest file and supports both agent skills — reusable instructions and resources for AI agents — and MCP server configurations, which connect agents to outside tools and services.

The pitch is simple: instead of every agent tool expecting a different folder layout and setup, there's now one shared format — build once, run anywhere. A minimal plugin.json needs only two fields, a schema reference and a plugin name, and the package structure includes a skills/ folder, an optional mcp.json file, and optional client-specific namespaces like com.example.client/. Vercel initiated the proposal, refined collaboratively with representatives from AWS, Anysphere (Cursor's maker), GitHub, Microsoft, OpenAI, and Vercel itself. At launch, the format is already supported by ChatGPT and Codex, Cursor, GitHub Copilot, Kiro, and VS Code.

• A minimal plugin.json needs just two fields: a schema and a plugin name
• Supported at launch by ChatGPT, Codex, Cursor, GitHub Copilot, Kiro, and VS Code
• Version 1.0.0 has no permission model, sandboxing, signature checks, or secrets mechanism yet — those are future work

The standard is governed by a technical steering committee with core maintainers from Amazon, Cursor, Microsoft, OpenAI, and Vercel — proposals and technical decisions are public, and participation is open to the wider community on GitHub. The standard's arrival is a logical next step given how fast the AI agent market is growing: French regulators found three companies already control over 84% of that market, and OpenAI itself just shipped its first-ever piece of hardware, a keyboard built for its Codex agent, back in July. A shared plugin format should cut down on friction between competing agent ecosystems right as the industry is still setting its standards.

None of this should be read as personalized investment advice.