VibeXP serves a single, team-agnostic MCP endpoint with its own OAuth 2.1 authorization server. Paste the URL into your client, approve once in the browser, and your AI can read and write your prompts, rules, memory, artifacts and feeds.
Most integrations start with generating a secret, pasting it into a config file and remembering to rotate it.
Every new editor or agent needs its own plugin, and the context it can reach differs from the last one.
Without a connection, you are the bridge, moving context between your notes and the chat by hand.
The client discovers VibeXP's authorization server, registers itself and opens a consent screen in your browser. Approve it and the client keeps a short-lived, audience-bound token that it refreshes on its own.
claude mcp add --transport http vibexp \ https://vibexp.example.com/mcp/v1/common
Search across every kind of knowledge, render prompts, read and write memories, artifacts and blueprints, post to feeds, and link resources to each other, all as MCP tools.
vibexp_io_search # find by meaning vibexp_io_render_prompt # fill a template vibexp_io_create_memory # write back vibexp_io_post_to_feed # report progress
An open standard for connecting AI tools to external data and tools. VibeXP implements it so any MCP client can reach your knowledge base through one endpoint.
In the docs ↗No. The MCP endpoint accepts only OAuth bearer tokens issued by the connect flow, and rejects API keys. API keys are for the CLI, the REST API and your own scripts.
In the docs ↗Then it cannot connect to the VibeXP MCP endpoint today. Clients that implement auto-discovery, dynamic client registration and PKCE connect by pasting the URL.
In the docs ↗Stars are how the next person tired of re-explaining their project to an AI finds this. It is the whole of the distribution: no ad budget, no sales team, no hosted plan to upsell.
Free, open source and self-hosted. One command to run it, one to connect your AI tools, and everything else in the docs.