brain mcp

Working with an agent that lacks context of your personal preferences, past decisions, and project history leads to repetitive explanations and shallow assistance. This skill bridges the gap by connecting your agent to your Obsidian vault via an MCP server, allowing it to treat your notes as its primary source of truth.

What it does

• Load identity - retrieves your specific profile, work preferences, and project goals to tailor every interaction.
• Contextual retrieval queries your vault for specific facts, decisions, or project details using hybrid search.
• Activity tracking summarizes recent work by analyzing note modifications and project progress over specific timeframes.
• Memory persistence saves new preferences, constraints, or durable facts back to your vault with explicit consent.
• Document capture generates meeting notes, research summaries, and project plans directly into your Captures/ folder.
• Graph exploration discovers relationships between notes and concepts to find related work or dependencies.

How it works

1. Initialize the connection to your brain-mcp server to verify the agent can access your Markdown files.
2. Retrieve context using tools like brain_identity or brain_context to ground the current task in your existing documentation.
3. Synthesize answers by citing specific note IDs and vault paths so you can verify the sources of the agent's suggestions.
4. Update records via brain_remember or brain_capture to ensure new insights or changes are stored for future sessions.

Frameworks & tools

This skill requires the brain-mcp server and an Obsidian vault containing plain Markdown files. It uses hybrid keyword and semantic search to interface with local file structures.

Why this beats prompting it yourself

Manual context injection is limited by token windows and requires you to constantly copy-paste your own notes. This skill automates the retrieval of exactly what is relevant, ensuring the agent "remembers" your feedback and preferences without manual reminders.

Use cases

• Project planning by referencing previous decision records and progress logs to suggest next steps.
• Onboarding agents to your personal coding style and workflow constraints from a central identity note.
• Recalling history to answer questions about why specific architectural choices were made months ago.
• Automated journaling by capturing session logs and durable facts directly into your second brain.

Known limitations

The agent cannot run the initial server installation and requires the vault to be pre-existing. It only has write access to designated Notes/ and Captures/ folders.