clawdi-ai/store/gbrain

GBrain — Shared Knowledge Brain

GBrain stores knowledge that persists across sessions and is shared between all agents. It's your long-term memory.

IMPORTANT: Search Automatically

When the user asks about past decisions, people, partnerships, meetings, or any context — ALWAYS search gbrain first before saying "I don't know."

MCP Tools (via mcporter)

# Search the brain (USE THIS FIRST)
mcporter call gbrain.search '{"query":"flashbots partnership status"}'
 
# Read a specific page
mcporter call gbrain.get_page '{"slug":"flashbots-phala-qa"}'
 
# List all pages
mcporter call gbrain.list_pages '{}'
 
# Save new knowledge
mcporter call gbrain.put_page '{"slug":"meeting-2026-04-17","content":"---\ntitle: Meeting Notes\ntype: concept\ntags: [meeting]\n---\n\n# Meeting Notes\n\n..."}'
 
# Delete a page
mcporter call gbrain.delete_page '{"slug":"old-page"}'

CLI Commands

# Search
gbrain search "query"
 
# Import a directory of markdown files
gbrain import /path/to/docs --no-embed
gbrain embed --stale
 
# List pages
gbrain list
 
# Health check
gbrain doctor --json
 
# Sync from git repo
gbrain sync --repo /root/brain --no-pull --no-embed
gbrain embed --stale

What's in the Brain

The brain contains knowledge from all agent workspaces:

• Meeting notes and transcripts
• Partnership context (Flashbots, etc.)
• Product decisions and PRDs
• User issues and resolutions
• Marketing plans and campaigns
• Sales pipeline and CRM context
• Technical docs and architecture decisions

When to Write to the Brain

Save important information that should persist:

• After a significant meeting or decision
• New partnership or relationship context
• Product strategy changes
• User feedback patterns
• Anything you'd want to remember next week

Page Format

---
title: Page Title
type: concept|source|person|company|project|media
tags: [tag1, tag2]
---
 
# Page Title
 
Content here...

Environment

Requires in .gbrain.env:

export OPENAI_API_KEY=<s2a-admin-key>
export OPENAI_BASE_URL=<s2a-proxy-url>/v1

Daily Sync

A daily cron job runs at 3:15 AM Pacific:

1. gbrain sync — pulls new docs
2. gbrain embed --stale — embeds new/changed pages
3. gbrain doctor --json — health check