beyin-mcp
A tool server running over the Model Context Protocol that searches a 247-file knowledge base in half a second and always returns the result together with its source path.
There are two ways to give an AI agent access to company knowledge: stuff everything into the context every time, or hand the agent a tool it can search. The first is slow and expensive, and the agent still cannot tell you where it learned something. beyin-mcp builds the second: it indexes the knowledge base, returns the files that match the question, and puts the source path next to every result. The answer becomes auditable, and whether it was invented can be checked against the source.
Problem
Two things go wrong when AI agents are pointed at company data. First, cost: pouring the whole knowledge base into every conversation is expensive, and past a certain size it no longer fits. Second, trust: if an agent cannot say where an answer came from, that answer cannot be audited, and an answer that cannot be audited cannot be used for real work.
How it works
On start-up the server scans the knowledge base and builds an in-memory index. When an agent calls the search tool, the query is matched against that index and the most relevant files come back with the sections that matched, each labelled with the file it came from, so the agent can cite its source. Because the server speaks the standard MCP interface, it connects the same way to any client that supports the protocol.
Screens
Highlights
Answers with their source
Every search result says which file it came from. The agent's answer becomes verifiable, which is the precondition for using it at work.
Access without filling the context
The whole knowledge base is never loaded into the conversation; the agent searches for what it needs. Cost stops growing in step with the size of the knowledge base.
Standard protocol
It connects to any client that speaks the Model Context Protocol, so you are not tied to a single vendor or a single interface.
Half-second search
A search across 247 files returns in about 0.5 seconds, fast enough not to break the flow of a conversation.
Let's talk
Let's start with a free 30-minute call; just pick a time that suits you from the calendar.