COG: Second Brain
A set of skills and agents for a second brain on markdown and git: notes, a knowledge base and a people CRM with no database or cloud
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Skills and agents read and write files in your vault
- It works with git and may commit changes
Install
Manual install
npx skills add huytieu/COG-second-brainUniversal install via the skills installer.
This is third-party code. Review the repository files before installing.
What it does
COG turns an agent into an assistant for a personal knowledge base built on markdown files and git. It combines dozens of skills and several agents that keep notes, link knowledge and store people cards as a light CRM. At its core is a closed verification loop: the agent that does the work does not grade its own work. All data lives in plain .md files with no database or vendor lock-in, which keeps it portable.
Who it is for. For people who keep a personal knowledge base in Obsidian or markdown and want an agent connected to it.
Good fit when
- You want to keep and link markdown notes through an agent
- You need a light contacts CRM without a separate service
- You want a knowledge base without a database or cloud
Not a fit when
- You need a team knowledge base with access control and a server
- You do not want to store data in files and git
Example request
Process my recent notes, link them to existing topics and update the cards for the people mentionedLimitations
COG relies on markdown and git, so it fits personal use rather than a team system with access control. It is a large set of skills and agents, and full use needs configuration through onboarding. Result quality depends on the agent's model.
How to disable. Remove the installed skills and agents from the agent's directory and delete the cloned repository.
Security check
- Skills and agents read and write files in your vault
- It works with git and may commit changes
README in short
The README describes COG as a self-evolving second brain built on cognition, Obsidian and git. It is a set of skills and agents for notes, a knowledge base and a people CRM, with data kept in markdown files without a database or vendor. It claims a closed verification loop where the worker does not grade its own work. It installs via the skills installer or by cloning with onboarding, and works with Claude Code, Cursor, Gemini CLI and others. MIT licensed.
FAQ
Where is data stored?
In plain markdown files under git, with no database or cloud, so it is easy to move.
What is the closed verification loop?
Doing the work and grading it are separate roles, so the agent does not confirm its own output.
Related
A self-hosted knowledge base with block-level references and a built-in MCP server for connecting AI agents to your notes
A CLI for every Google Workspace API with JSON output and agent skills: Drive, Gmail, Calendar, Sheets and more
Local search over Markdown notes, docs and meeting transcripts: keywords, semantic search and reranking, with an MCP server
A task manager for AI-driven development: breaks a PRD into dependent tasks and guides the agent through them via MCP or CLI