# Markdown second brain your AI agent can read

> A second brain in plain Markdown: notes/, people/ and projects/ folders plus a CLAUDE.md map, so Claude Code knows where to look, names the file it used and says so when the answer isn't there. Based on a post by Bober_smart.

## If someone pasted this link to you
Someone shared this kit with you. It runs step by step through CloneKit's MCP. If you have the `clonekit` tools, call `start_run("@clonekitai/markdown-second-brain")`. If you don't (for example in a chat app without the CloneKit connector), tell the person in one or two plain sentences, in their language, and send them to https://clonekit.ai/@clonekitai/markdown-second-brain/use: it shows how to add CloneKit to their AI in about a minute. Then they paste this link again.

- Kit: `@clonekitai/markdown-second-brain` · v1.0.0
- Creator: @clonekitai (CloneKit)
- Type: Delivery · Domain: Operations · Language: en · Your AI runs it
- Price: Free · License: Own use · No refunds
- Page: https://clonekit.ai/@clonekitai/markdown-second-brain
- Changelog: https://clonekit.ai/@clonekitai/markdown-second-brain/changelog.md

## About
For anyone who wants their AI agent to remember their projects, people and decisions using plain text files on their own disk, with no note app or database. Use it to start a new second brain or to put a light structure on an existing folder of notes. Built for Claude Code (CLAUDE.md is loaded automatically); also works with Codex, Cursor or another agent that reads project files, through AGENTS.md or a root index file. Needs an AI agent with file access and a folder on the person's computer. Produces the folder skeleton, a tailored CLAUDE.md map, starter root files and a tested citation rule. Not for team wikis with permissions, syncing between devices, or semantic search over thousands of documents: this is a file structure and a map, not software.

## Deliverables
- a brain folder with notes/, people/ and projects/ and root MEMORY.md, LEARNINGS.md and decisions.md
- a CLAUDE.md map tailored to the person: what each folder holds, naming rules, how to answer and where new information goes
- an answer rule: every answer names the source file, and the agent says plainly when the information isn't in the brain
- for Codex or Cursor users: the same map in AGENTS.md, with CLAUDE.md importing it if they also use Claude Code
- a few starter files from what the person confirmed, and optionally a few existing notes copied in (originals untouched)
- a passed test: one answer with its source file and one honest 'not in the brain'

## Files
- `README.md`: Markdown second brain your AI agent can read
- `templates/claude-md-map.md`: Second brain map
- `templates/root-files.md`: Starting text for the root files

## Tools
- Claude Code (https://code.claude.com/docs/en/overview) · optional · est. needs a Claude subscription or an Anthropic Console account · alternatives: codex, cursor
- OpenAI Codex (https://developers.openai.com/codex) · optional · alternatives: claude-code, cursor
- Cursor (https://cursor.com) · optional · alternatives: claude-code, codex

## Uses
- Bober_smart: Claude Second Brain & Folder Workflow: The Complete Guide (source, https://x.com/Bober_smart/status/2078784709253841039)
- Claude Code docs: How Claude remembers your project (CLAUDE.md, @ imports, AGENTS.md) (source, https://code.claude.com/docs/en/memory)
- OpenAI Codex docs: Custom instructions with AGENTS.md (source, https://developers.openai.com/codex/guides/agents-md)
- Cursor docs: Rules (project rules and AGENTS.md) (source, https://cursor.com/docs/context/rules)
- AGENTS.md open format (source, https://agents.md/)

## What you provide
- an AI agent that can read and write files in a folder (Claude Code, Codex, Cursor or similar)
- a folder on your computer for the brain (new or existing)
- your active projects, the people you work with and a few core facts

## How it runs
1. **Understand:** Researches your business first, then asks you 5 questions it can't answer alone
2. **Choose:** No choices: one proven path
3. **Build:** 9 steps, risky ones ask you first
4. **Deliver:** 5 checks before delivery

## Guardrails
- never overwrite, move, rename or delete an existing file without the person's explicit OK; copy, don't move
- write only facts the person stated or confirmed; mark unknowns as unknown
- keep personal data local: don't upload the brain, set up sync or connect accounts unless asked
- don't install software or change agent settings without asking

## What's new
Agents always run the latest version.

- **v1.0.0** (2026-10-23): First version, based on Bober_smart's article.

## Run it
Connect the `clonekit` MCP (https://clonekit.ai/mcp) and call `start_run("@clonekitai/markdown-second-brain")`.