Skill Claude

Build Brain

build-brain/

2 views 0 downloads

Turn a skill or a pile of notes into an AI-optimized second brain: a lean map, small tagged references and JSON navigation, plus a before and after token report.

Build Brain: Brain Lab skill for Claude Code

Turn a skill or a pile of raw notes into an AI-optimized second brain: a lean map + atomized references + JSON nav.

Why use it

One giant 6,000-token document gets loaded in full to answer a one-line question.
Knowledge nobody can navigate is knowledge nobody uses, including the agent.
Context windows are budget; a dump spends it all on the first turn.

What it does

Analyzes the source, then restructures it into a lean SKILL.md map plus atomized, tagged references.
Generates JSON navigation maps (index / tags / links / manifest) so an agent can find one note instead of reading everything.
Triple-checks the result and writes a before/after improvement report with the token numbers.

Benefits

Slash your context bill. Load a small map, open one note, not a 6,000-token dump.
Recall gets cheap, so the agent actually uses the knowledge you wrote.
Scales with you. Knowledge stays navigable as it grows.
See the win. A before/after report with the real token numbers.

Install

Download this skill and copy the build-brain folder to .claude/skills/build-brain/ in your project, or to ~/.claude/skills/build-brain/ to use it in every project. Claude Code finds it on its own; you can also call it with /build-brain.

Part of the Context Engineering bundle. To get the whole set as a Claude Code plugin: /plugin marketplace add endorphin-ai/hasbrains-agent-kit, then /plugin install context-engineering@hasbrains-agent-kit.

Source: endorphin-ai/hasbrains-agent-kit. License: MIT.

Files (5)

# build-brain

> Turn a skill or a pile of raw notes into an AI-optimized second brain: a lean map + atomized references + JSON nav.

**Plugin:** `context-engineering` — `/plugin install context-engineering@hasbrains-agent-kit`

## Why use it

- One giant 6,000-token document gets loaded in full to answer a one-line question.
- Knowledge nobody can navigate is knowledge nobody uses — including the agent.
- Context windows are budget; a dump spends it all on the first turn.

## What it does

- Analyzes the source, then restructures it into a lean `SKILL.md` **map** plus atomized, tagged references.
- Generates JSON navigation maps (index / tags / links / manifest) so an agent can find one note instead of reading everything.
- Triple-checks the result and writes a before/after improvement report with the token numbers.

## Benefits

- **Slash your context bill.** Load a small map, open one note — not a 6,000-token dump.
- **Recall gets cheap**, so the agent actually uses the knowledge you wrote.
- **Scales with you.** Knowledge stays navigable as it grows.
- **See the win.** A before/after report with the real token numbers.

---

Made by **HasBrains** — https://hasbrains.com/