Layer System

L1/L2/L3 distribution layers for packaging and privacy control

Mega Brain organizes content into three distribution layers (L1, L2, L3) that control what is shared publicly, distributed to premium users, or kept private.

Quick Reference

Layer	Name	Git Status	Distribution	Who Uses It
L1	Community	Tracked	npm package (public)	Everyone — open source core
L2	Premium	Tracked	Private repo (gated)	Paying users
L3	Personal	Gitignored	Local only / personal backup	Individual user only

Layer 1: Community (npm package)

Purpose

The open-source engine that powers Mega Brain. Distributed via npm as mega-brain-ai. Anyone can install and use this to build their own knowledge management system.

Git status: Fully tracked
Distribution: Public — npm publish
Who uses it: Open source community, all users as the foundation layer

What Belongs in L1

Core Engine

• core/ - Processing engine
• core/tasks/ - Task definitions
• core/workflows/ - Workflow templates
• core/schemas/ - JSON schemas
• core/intelligence/ - Intelligence scripts

CLI & Tooling

• bin/ - CLI binaries
• Setup wizard
• Validation scripts
• Pre-publish gates

Integration

• .claude/ - Claude Code integration
• .claude/hooks/ - Lifecycle hooks
• .claude/commands/ - Slash commands
• .claude/skills/ - Skill system
• .claude/rules/ - Operational rules

Templates

• agents/_templates/ - Agent templates
• agents/conclave/ - Deliberation agents
• docs/ - Documentation
• .gitkeep files (structure markers)

Real Examples

core/                          → L1 (Core engine)
core/tasks/                    → L1 (Core engine)
core/intelligence/audit_layers.py → L1 (Core engine)
bin/                           → L1 (Core engine)
.claude/                       → L1 (Core engine)
.claude/rules/RULE-GROUP-1.md  → L1 (Core engine)
agents/conclave/               → L1 (Core engine)
docs/                          → L1 (Core engine)
inbox/.gitkeep                 → L1 (Empty structure marker)

Key rule: .gitkeep files are always L1, regardless of directory. They mark where populated content will live but contain no personal data.

Layer 2: Premium (populated content)

Purpose

Content generated through the Mega Brain pipeline — actual knowledge, agent personalities, dossiers, playbooks. Requires a paid subscription or private repository access.

Git status: Tracked in private repo (gitignored in public L1 repo)
Distribution: Private — premium repository or direct sync
Who uses it: Paying users who have run the pipeline and built their knowledge base

What Belongs in L2

Agents

• agents/minds/ (populated)
• agents/cargo/ (populated)
• Agent personalities and memories
• Agent DNA configurations

Knowledge

• knowledge/dossiers/
• knowledge/playbooks/
• knowledge/dna/
• knowledge/sources/

Artifacts

• artifacts/insights/
• artifacts/chunks/
• artifacts/extractions/
• Populated state files

Processed Content

• Dossiers with expert knowledge
• Extracted DNA schemas
• Synthesized narratives
• Theme compilations

Real Examples

agents/minds/                        → L2 (Premium content — when populated)
agents/cargo/                        → L2 (Premium content — when populated)
knowledge/dossiers/persons/          → L2 (Premium content)
knowledge/dossiers/themes/           → L2 (Premium content)
knowledge/playbooks/                 → L2 (Premium content)
knowledge/dna/                       → L2 (Premium content)
knowledge/sources/                   → L2 (Premium content)
artifacts/insights/                  → L2 (Premium content)
artifacts/chunks/                    → L2 (Premium content)

Key rule: L2 is a superset of L1. An L2 distribution includes all L1 content plus the populated knowledge content.

Layer 3: Personal (never distributed)

Purpose

User-generated content specific to one person’s workflow — raw source materials, processing logs, session history, company-specific data. Never leaves the local machine except in personal backups.

Git status: Gitignored (not committed to any shared repo)
Distribution: Local backup only — never shared
Who uses it: The individual user only — this is their private data

What Belongs in L3

Raw Materials

• inbox/ - All input materials
• Videos, PDFs, transcripts
• Course materials
• Podcast downloads

Logs & History

• logs/ - Processing logs
• .claude/sessions/ - Session history
• Execution reports
• Audit trails

State & Runtime

• .claude/mission-control/
• Session state files
• Runtime configuration
• Processing checkpoints

Company Data

• agents/sua-empresa/
• Company-specific agents
• Internal processes
• Proprietary workflows

Real Examples

inbox/                         → L3 (Personal data)
inbox/alex-hormozi/            → L3 (Personal data)
inbox/my-video-transcript.txt  → L3 (Personal data)
logs/                          → L3 (Personal data)
logs/batches/BATCH-001.md      → L3 (Personal data)
.claude/sessions/              → L3 (Personal data)
.claude/mission-control/       → L3 (Personal data)
agents/sua-empresa/            → L3 (Personal data)

Key rule: L3 content is the “input” and “runtime state” of your system. Valuable to you personally but meaningless (or sensitive) to others.

Special Categories

NEVER — Secrets and Credentials

Purpose: Files that must never be committed to any git repository.

What belongs:

• .env, .env.local - Environment variables
• credentials.json, token.json - API credentials
• .mcp.json - MCP configuration with tokens
• *.key, *.pem - Certificate files

.env                           → NEVER (Secrets/sensitive config)
.env.local                     → NEVER (Secrets/sensitive config)
.mcp.json                      → NEVER (Secrets/sensitive config)
credentials.json               → NEVER (Secrets/sensitive config)
token.json                     → NEVER (Secrets/sensitive config)

DELETE — Obsolete Content

Purpose: Files superseded by newer implementations.

Run python3 core/intelligence/audit_layers.py to identify delete candidates.

Classification Flowchart

1. Check NEVER

   Does it contain API keys, tokens, or credentials?

   YES → NEVER (add to .gitignore)

2. Check L3

   Is it user-generated (inbox, logs, sessions, company data)?

   YES → L3 (gitignored)

3. Check L2

   Is it populated knowledge (dossiers, playbooks, DNA, artifacts)?

   YES → L2 (private repo)

4. Check .gitkeep

   Is it a .gitkeep file anywhere?

   YES → L1 (structure marker)

5. Default to L1

   Is it core engine, CLI, integration, or templates?

   YES → L1 (public repo)

Distribution Matrix

Content Type	L1 (Community)	L2 (Premium)	L3 (Personal)
Core engine	✅	✅	✅
CLI tools	✅	✅	✅
Agent templates	✅	✅	✅
Empty structure	✅	✅	✅
Populated agents	❌	✅	✅
Knowledge base	❌	✅	✅
Processing artifacts	❌	✅	✅
Raw materials	❌	❌	✅
Logs & sessions	❌	❌	✅
Company data	❌	❌	✅
API keys	❌	❌	❌

Community vs Pro Features

Feature	Community (L1)	Pro (L2)
CLI and setup wizard	✅	✅
Core engine and templates	✅	✅
Skills and hooks	✅	✅
Agent templates and examples	✅	✅
Populated knowledge base	❌	✅
Mind clone agents	❌	✅
Pipeline processing	❌	✅
Council / Conclave	❌	✅

Validation

Verify layer classification before publishing:

# Check that only L1 content is in the package
npm run validate:layers

# Full pre-publish gate (secrets scan + layer validation)
node bin/pre-publish-gate.js

# Programmatic audit with Python
python3 core/intelligence/audit_layers.py

Layer Management Best Practices

1. Separate Concerns

   Keep L1 (engine), L2 (content), and L3 (personal) clearly separated

2. Never Mix

   Don’t commit L3 content to L1/L2 repos. Don’t include secrets anywhere.

3. Validate Before Publish

   Always run validation before pushing to public or premium repos

4. Document Decisions

   When adding new content, document which layer it belongs to and why

Next Steps

Architecture

Understand the full system architecture

Knowledge Pipeline

Learn how L2 content is generated