Vercel Agent Plugins 1.0.0 is an open, vendor-neutral standard specification for packaging AI agent extensions, designed for developers who build plugins for ChatGPT, Cursor, GitHub Copilot, Kiro, VS Code and Codex without rewriting for each client.
Core Features
- A minimal plugin.json manifest (just $schema and name) plus fixed locations for skills/ and mcp.json, so one package works across every compatible client.
- Bundles two portable component types: Agent Skills (SKILL.md instruction files) and MCP server configurations (stdio, Streamable HTTP, legacy SSE).
- Independent component validation with failure isolation, so a broken skill does not disable the MCP server in the same plugin.
- Namespaced client extensions (com.example.client/) that other clients safely ignore, keeping the shared format small and easy to implement.
Use Cases
- Plugin authors who want to ship one package to six agent clients instead of maintaining six separate repos.
- Agent-client builders implementing a conformance checklist for discovery and loading.
- Teams standardizing how internal tools and MCP servers reach their AI assistants.
Pricing
Agent Plugins is a free, openly licensed specification with public governance. There is no paid plan; adoption is the only cost. Vercel's v0 CLI can install provider skills automatically when you add a Marketplace integration.
Our Take
Best for teams shipping agent extensions across multiple AI coding clients. The trade-off is scope: v1.0.0 covers packaging only, so permissions, sandboxing and signing are explicitly deferred and you should still treat any plugin as code execution. Co-designed by AWS, OpenAI, Microsoft, Cursor and Vercel, it is the broadest day-one standard yet. See more in our AI Programming development directory.
