BrainTrain

MCP Memory Server for Claude Code "Your AI never forgets the journey"

Concept

BrainTrain is a memory system for Claude Code based on a train metaphor:

• Locomotive - Master session with base knowledge (DevStandards, architecture, code style)
• Wagons - Work sessions, linked in a chain
• Train - Entire project (locomotive + wagons)

┌─────────────┐    ┌─────────┐    ┌─────────┐    ┌─────────┐
│ LOCOMOTIVE  │───▶│ Wagon 1 │───▶│ Wagon 2 │───▶│ Wagon N │
│  (Master)   │    │         │    │         │    │         │
├─────────────┤    ├─────────┤    ├─────────┤    ├─────────┤
│ • DevStd    │    │ Context │    │ Context │    │ Context │
│ • Architect │    │ Locomot.│    │ Locomot.│    │ Locomot.│
│ • Style     │    │    +    │    │    +    │    │    +    │
│             │    │  own    │    │ Wagon 1 │    │ Wagon.. │
└─────────────┘    └─────────┘    └─────────┘    └─────────┘

Installation

claude mcp add braintrain -- npx -y braintrain

Or via GitHub:

claude mcp add braintrain -- npx -y github:temrjan/braintrain

Usage

Start a new project

bt_locomotive_create("myproject", "My awesome project")
bt_locomotive_learn("backend", "FastAPI, PostgreSQL, structure...")
bt_locomotive_learn("frontend", "React 19, Vite, components...")
bt_locomotive_learn("devstandards", "PM2, releases/current, ...")

Work session

bt_wagon_start("Add product filter")
bt_recall()  # Get locomotive context

[Work on the task...]

bt_wagon_save(
  summary: "Added category filter to admin",
  decisions: ["Used useState for filterCategoryId"]
)

Next session

bt_wagon_start()
bt_recall()

# Context:
# - Locomotive: DevStandards, architecture, style
# - Wagon 1: "Added category filter..."

# Ready to continue!

MCP Tools

Locomotive

| Tool | Description | |------|-------------| | bt_locomotive_create | Create new locomotive for a project | | bt_locomotive_learn | Add knowledge to locomotive | | bt_locomotive_update | Update standards or description | | bt_locomotive_reset | Reset (archives current first) | | bt_locomotive_clear | Clear completely without archiving |

Wagon

| Tool | Description | |------|-------------| | bt_wagon_start | Start new wagon (work session) | | bt_wagon_save | Save and close current wagon | | bt_wagon_clear | Clear old wagons (keep recent N) |

Context

| Tool | Description | |------|-------------| | bt_recall | Get full context (locomotive + wagons) | | bt_status | Get train status |

Archive

| Tool | Description | |------|-------------| | bt_archive | Archive current project | | bt_archive_list | List all archives | | bt_archive_restore | Restore from archive |

Data Storage

~/.braintrain/
├── config.json
├── active/
│   └── {project}/
│       ├── locomotive.json
│       └── wagons/
│           ├── 001.json
│           ├── 002.json
│           └── current.json
└── archive/
    └── {project}/
        └── {timestamp}/

License

MIT

"Train your brain, never lose the chain"