# Second Brain Starter — a project by Alejandro Marco

> I turned the second-brain system I use personally into an open-source starter, so anyone can adapt the same Markdown structure, agent guidance and practical workflows to their own life.

- **Type:** Open-source starter
- **Role:** Product, system design and engineering
- **When:** 2026
- **Human page:** [https://alejandromarco.com/projects/second-brain-starter](https://alejandromarco.com/projects/second-brain-starter)

## Why I made it

I wanted to share a second brain that agents can genuinely work with while the knowledge stays legible, portable and under the control of the person using it.

## The project

This starter is a shareable version of the second-brain system I use myself. I extracted the useful patterns into something another person can understand in an afternoon, rather than publishing a finished vault that only makes sense with my history inside it.

In my own setup, OpenClaw uses it from a Mac mini when I want persistent context. I also work against the repository directly with Codex for more focused sessions, including work related to my master’s. Both operate on the same plain Markdown instead of separate copies of my knowledge.

The reusable version keeps that portability. Obsidian is optional, GitHub sync is private by default, and agent-specific personality, credentials and permissions stay outside the vault. The included CV tool makes the model tangible: canonical career facts can become tailored HTML and PDF documents, including a public version without email, without turning generated wording into a new source of truth.

## What I built

### The system I use, made reusable

I extracted the structure and working rules from my own second brain, then added fictional connected examples and guided setup so someone else can understand it, personalize it and remove the demo safely.

### One source of context across agents

I use the vault with OpenClaw running on my Mac mini, and sometimes open the same repository directly in Codex for focused work, including projects from my master’s. Plain Markdown keeps that context portable between both.

### Enough structure without a second job

I separate capture, journals, sources, ideas, projects, areas, people and decisions by what each note does. The model stays intentionally small, and uncertain material can remain in the inbox instead of being forced into a taxonomy.

### Agent workflows with human control

I included repository-level instructions and local skills for guided onboarding, safe knowledge capture and CV generation. They preserve provenance, forbid invented personal facts and require explicit approval before publishing, deleting or changing external systems.

### Canonical facts become useful artifacts

I added a CV workflow that keeps career facts in one Markdown source, creates tailored HTML and PDF versions, supports a public export without email, and asks for visual and factual verification before anything is shared.

## Scope and limitations

The repository ships with a connected fictional example so the system can be understood before it is personalized. It is a starter rather than a hosted sync service, and its ignored private folder is a safety boundary rather than encryption. I recommend keeping personal copies private, reviewing every agent skill like code and inspecting diffs before committing.

## Links

- [Use the starter](https://github.com/AlejandroMI/second-brain-starter)