Browse / Second brain / brain-maxed

brain-maxed

octopodas

Deterministic second brain for your Claude Code workspace — index memories, skills, MCP servers, plugins & routines, retrieve by scoring the index without opening files, and render it as a ring/force graph.

Open octopodas/brain-maxed → Does not match the four rules.

Built for one agent

ships CLAUDE.md and nothing for other agents.

In its own words

Aboutoctopodas/brain-maxed

Zero deps, one file.

README.md

A deterministic second brain for your agentic workspace. Plain code answers "where is that fact?" — the model only sees the final evidence. One file (brain.js), no dependencies, no server.

It indexes your memories (~/.claude/memory, ./memory), all skills, MCP servers, plugins, connectors, and routines into INDEX.md — one line per source — then retrieves by scoring that index without opening files.

Requirements

Node.js. Nothing else.

Commands

All commands run from this folder.

Ask a question

Strips your question to keywords, scores every index line without opening any files, opens the single best file, and prints the single best section — plus where it came from:

If the best section is just a pointer ([[slug]] or a .md link), it follows it once automatically.

Lead the question with list (e.g. node brain.js "list recent plans") to get every matching index entry, newest-first, instead of one section.

Store a memory

Writes memory/deploy-cadence.md and its index line in one step. --name is optional — without it the slug is built from the first keywords. Re-using a name overwrites (that's how you update a fact).

Rebuild the index

Rescans everything on disk. Run after adding/removing skills, MCP servers, or memory files. Safe to run anytime — the index is always regenerated from reality, so it can't drift.

The repository's own description, then 32% of its README. Read the rest on GitHub ↗

Read from the repository's git tree on 2026-08-17.