Show PH: Plurality's Open Context MCP - take your AI memory to any tool, anywhere
Hey PH 👋
I want to be upfront about something before I share what we built.
We've been building AI Context Flow - inline context injection via a browser extension that takes your personal context into AI tools as you use them. It's been working well, but it always had a ceiling.
Inline injection can only go so far. There are corner cases where you need deeper context. And the extension is browser-only. What about desktop apps, CLI tools, local models, mobile?
You shouldn't have to keep re-explaining yourself to every AI in every environment.
So we built the next layer.
Introducing Plurality's Open Context MCP: a way to take your context to any AI application, anywhere.
If you've been following the MCP wave, think of it like the USB-C of AI. A universal, standardized way to stitch systems together.
The browser extension was a patch. This is infrastructure.
Here's what makes it different at the architecture level:
The AI pulls what it needs, when it needs it (via semantic search across your memory)
It writes back. Conversations, decisions, and new context get saved from within any chat
It works outside the browser entirely (IDE, CLI, local models, anywhere)
Every tool connects to the same memory layer i.e. context built in Cursor shows up in Claude Code shows up in ChatGPT
You add the MCP server once. After that, you're not pasting context into AI tools anymore. You're giving them access to a memory layer they can read from and write to. One that you own, and that persists across every session.
What's supported today: ChatGPT, Claude Desktop, Claude Code, GitHub Copilot, Cursor, Windsurf, LM Studio, Lovable, Replit and any other client that speaks MCP over HTTP.
Setup guide with exact steps for every tool: https://plurality.network/blogs/connect-ai-context-flow-anywhere-using-mcp-servers/
Happy to answer questions about how the OAuth flow works, how memory buckets are structured, or anything else under the hood.
Replies