Agent Plugins Explained: How We Built the Context7 Plugin
Blog post from Upstash
Agent Plugins is a new open standard backed by Vercel, AWS, Cursor, GitHub, Microsoft, and OpenAI that packages agent extensions as a single installable directory, reducing the need for separate client-specific integrations for an MCP server. Context7, which supplies current version-specific library documentation to coding agents, has adopted the format with a manifest, an MCP configuration pointing to its remote server, and a skill defining when and how documentation should be retrieved. The specification uses fixed file locations rather than manifest-declared component paths and currently supports only skills and MCP servers, while commands, hooks, subagents, and persistent rules remain specific to individual clients. Because plugins cannot contain secrets or expand environment-variable placeholders in connection headers, the Context7 plugin uses standard MCP OAuth authorization instead of API keys. The format remains a working draft and does not yet address distribution, sandboxing, permissions, or publisher trust, while Context7’s older client-specific plugins remain available for users needing unsupported features.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| MCP | 19 | 1,562 | 186 | 99 | -80% |
| AI Coding Assistant | 3 | 276 | 77 | 47 | -83% |
| LLM | 1 | 1,189 | 251 | 109 | -83% |
| Secrets Management | 1 | 584 | 99 | 52 | -76% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.