Second Brain With MCP: Connect AI to the Record You Keep
An MCP connection lets a supported AI assistant read and update your second brain during a session, with clear permissions over what it can change. Here is how that works and what to keep in a shared project record.
Updated 4 min read
An MCP (Model Context Protocol) connection gives a supported AI assistant a direct way to read your second brain during a session, rather than relying on you to paste in relevant notes before every conversation. The assistant calls a tool, the second brain returns the content, and the session starts from real context.
The key word is "supported": not every note-taking or knowledge tool exposes an MCP server. A purpose-built tool like fenbs is designed for exactly this connection, keeping project work, decisions, and progress in a form every connected assistant can read and update within the access you grant.
What MCP does in this context
MCP is a standard that AI apps implement to call outside tools during a conversation. When you connect an MCP server to your AI app, the assistant gains a set of callable functions: read context, list tasks, add a note, record a decision.
This is different from giving an assistant a document to read at the start. An MCP connection is live: the assistant can call the tool at any point in the session to fetch fresh information, and it can write back to the same record when it completes work. The record stays current across sessions and across different AI apps.
For a broader view of tools that connect over MCP, the MCP apps guide covers what different kinds of servers can do and how to compare them.
What belongs in an AI-readable second brain
A second brain that an assistant will read needs to be structured differently from a personal note archive. An AI assistant cannot infer context from a folder of loosely connected files. It needs clear, short, purposeful records.
- Decisions and rules: what was agreed, who agreed it, and whether it still applies. An assistant should read these before taking any action on the project.
- Where we left off: the last session's end state, what is in progress, and what is waiting on a person.
- Tasks: what needs doing, its current state, and any plan for completing it.
- Lessons: what went wrong before, what to do differently, read before similar work.
- Source links: references the assistant needs to cite or check, not full documents.
Long documents, full email threads, and raw notes are better kept where they live. The second brain that an assistant reads should hold the extracted useful facts: what was decided, where work stands, what to watch out for.
Permissions matter more than storage
The most important question when connecting an AI to a second brain is not what the tool stores but what the assistant is allowed to do with it. Read-only access is appropriate for reference material. Write access should be limited to specific actions: adding a task note, recording a decision a person already made, saving progress.
In fenbs, the people decide and the assistants record. An assistant can add a task, write a lesson, or save a handoff. It cannot sign off on a decision or pre-approve work. Every change is in History. That separation matters: it means the record reflects what people chose, not what the assistant inferred.
Give every assistant the same starting point
fenbs keeps the project record that any connected AI app reads at the start of a session: Rules first, then Where we left off, then the current tasks and context notes. When the session ends, the assistant saves progress for the next one. Create a free account, add your project, and connect your first AI app to see the context it reads.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
Switch AI apps without rebuilding the context
The biggest benefit of an external second brain over an in-app memory feature is portability. If your Claude session ends and you open Cursor, both connect to the same fenbs project and read the same context. The work does not live in one app's memory.
Before switching apps, ask the current assistant to save its progress: which tasks moved, what was decided, and where the next session should start. Then open the next app and ask it to read the same project context. It will start from where the previous session left off.
This is the practical form of what the AI second brain guide describes as keeping knowledge outside any single model. The portability comes from the external record, not from the assistant itself.
When fenbs is not the right fit
fenbs keeps the work record for projects: tasks, decisions, lessons, and progress. It is not a personal note archive, a document library, or a tool for storing research across unrelated topics. If you need to connect an assistant to a large collection of reference documents, a knowledge base tool or a RAG-based system (retrieval-augmented generation, a way of searching documents to add context) is a better fit for that job.
How to do this with fenbs
- Create a free account and add a project. Write a short description of the goal and add the first tasks.
- Write the first decisions and rules: things the assistant must read before working on this project.
- Go to Integrations and connect your AI app over MCP. Follow the guide for your app.
- At the start of each session, ask the assistant to read the project context from fenbs: it reads Rules first, then Where we left off.
- Before ending a session, ask the assistant to save progress so the next session starts from the current state, regardless of which AI app opens next.