shodh-memory

---

---

For AI Agents — Claude, Cursor, GPT, LangChain, AutoGPT, robotic systems, or your custom agents.
Give them memory that persists across sessions, learns from experience, and runs entirely on your hardware.

---

We built this because AI agents forget everything between sessions. They make the same mistakes, ask the same questions, lose context constantly.

Shodh-Memory fixes that. It's a cognitive memory system—Hebbian learning, activation decay, semantic consolidation—packed into a single ~17MB binary that runs offline. Deploy on cloud, edge devices, or air-gapped systems.

Quick Start

Choose your platform:

Platform	Install	Documentation
Claude / Cursor	claude mcp add shodh-memory -- npx -y @shodh/memory-mcp	MCP Setup
Python	pip install shodh-memory	Python Docs
Rust	cargo add shodh-memory	Rust Docs
npm (MCP)	npx -y @shodh/memory-mcp	npm Docs

TUI Dashboard

shodh-tui

Keyboard shortcuts: Tab switch panels · j/k navigate · Enter select · / search · q quit

GTD Todo System

Built-in task management following GTD (Getting Things Done) methodology:

# Add todos with context, projects, and priorities
memory.add_todo("Fix authentication bug", project="Backend", priority="high", contexts=["@computer"])

# List by project or context
todos = memory.list_todos(project="Backend", status=["todo", "in_progress"])

# Complete tasks (auto-creates next occurrence for recurring)
memory.complete_todo("SHO-abc123")

MCP Tools for Claude/Cursor:

• add_todo — Create tasks with projects, contexts, priorities, due dates
• list_todos — Filter by status, project, context, due date
• complete_todo — Mark done, auto-advances recurring tasks
• add_project / list_projects — Organize work into projects

How It Works

Experiences flow through three tiers based on Cowan's working memory model:

Working Memory ──overflow──▶ Session Memory ──importance──▶ Long-Term Memory
   (100 items)                  (500 MB)                      (RocksDB)

Cognitive Processing:

• Hebbian learning — Co-retrieved memories form stronger connections
• Activation decay — Unused memories fade: A(t) = A₀ · e^(-λt)
• Long-term potentiation — Frequently-used connections become permanent
• Entity extraction — TinyBERT NER identifies people, orgs, locations
• Spreading activation — Queries activate related memories through the graph
• Memory replay — Important memories replay during maintenance (like sleep)

Claude / Cursor (MCP)

Quick Start: Full Setup

The MCP client connects to a shodh-memory server. Follow these steps:

Step 1: Start the server

Download from GitHub Releases or use Docker:

# Option A: Direct download (Linux/macOS)
curl -L https://github.com/varun29ankuS/shodh-memory/releases/latest/download/shodh-memory-linux-x64.tar.gz | tar -xz
./shodh-memory

# Option B: Docker
docker run -d -p 3030:3030 -e SHODH_HOST=0.0.0.0 -v shodh-data:/data roshera/shodh-memory

Wait for "Server ready!" message before proceeding.

Step 2: Generate an API key

The API key is locally generated — you create your own. This is for local client-server authentication, not a cloud service credential:

# Generate a random key
openssl rand -hex 32
# Example output: a1b2c3d4e5f6...

Set this key on your server via SHODH_DEV_API_KEY environment variable.

Step 3: Configure the MCP client

Claude Code (CLI):

claude mcp add shodh-memory -- npx -y @shodh/memory-mcp

Claude Desktop / Cursor config:

{
  "mcpServers": {
    "shodh-memory": {
      "command": "npx",
      "args": ["-y", "@shodh/memory-mcp"],
      "env": {
        "SHODH_API_KEY": "your-generated-key-from-step-2"
      }
    }
  }
}

Step 4: Verify connection

curl http://localhost:3030/health
# Should return: {"status":"ok"}

Key MCP Tools:

• remember — Store memories with types (Observation, Decision, Learning, etc.)
• recall — Semantic/associative/hybrid search across memories
• proactive_context — Auto-surface relevant memories for current context
• add_todo / list_todos — GTD task management
• context_summary — Quick overview of recent learnings and decisions

Config file locations:

Editor	Path
Claude Desktop (macOS)	~/Library/Application Support/Claude/claude_desktop_config.json
Claude Desktop (Windows)	%APPDATA%Claudeclaude_desktop_config.json
Cursor	~/.cursor/mcp.json

Python

pip install shodh-memory

from shodh_memory import Memory

memory = Memory(storage_path="./my_data")
memory.remember("User prefers dark mode", memory_type="Decision")
results = memory.recall("user preferences", limit=5)

Full Python documentation →

Rust

[dependencies]
shodh-memory = "0.1"

use shodh_memory::{MemorySystem, MemoryConfig};

let memory = MemorySystem::new(MemoryConfig::default())?;
memory.remember("user-1", "User prefers dark mode", MemoryType::Decision, vec![])?;
let results = memory.recall("user-1", "user preferences", 5)?;

Full Rust documentation →

REST API

The server exposes a REST API on http://localhost:3030. All /api/* endpoints require the X-API-Key header.

Core Memory

Method	Endpoint	Description
POST	/api/remember	Store a memory
POST	/api/remember/batch	Store multiple memories
POST	/api/recall	Semantic search
POST	/api/recall/tags	Search by tags
POST	/api/proactive_context	Context-aware retrieval
POST	/api/context_summary	Get condensed summary
GET	/api/memory/{id}	Get memory by ID
DELETE	/api/memory/{id}	Delete memory
POST	/api/memories	List with filters
POST	/api/reinforce	Hebbian feedback

Todos

Method	Endpoint	Description
POST	/api/todos	List todos
POST	/api/todos/add	Create todo
POST	/api/todos/update	Update todo
POST	/api/todos/complete	Mark complete
POST	/api/todos/delete	Delete todo
GET	/api/todos/{id}	Get todo by ID
GET	/api/todos/{id}/subtasks	List subtasks
POST	/api/todos/stats	Get statistics

Projects

Method	Endpoint	Description
GET	/api/projects	List projects
POST	/api/projects/add	Create project
GET	/api/projects/{id}	Get project by ID
POST	/api/projects/delete	Delete project

Health

Method	Endpoint	Description
GET	/health	Health check
GET	/metrics	Prometheus metrics
GET	/api/context/status	Context window status

curl -X POST http://localhost:3030/api/remember 
  -H "Content-Type: application/json" 
  -H "X-API-Key: your-api-key" 
  -d '{
    "user_id": "user-1",
    "content": "User prefers dark mode",
    "memory_type": "Decision",
    "tags": ["preferences", "ui"]
  }'

curl -X POST http://localhost:3030/api/recall 
  -H "Content-Type: application/json" 
  -H "X-API-Key: your-api-key" 
  -d '{
    "user_id": "user-1",
    "query": "user preferences",
    "limit": 5
  }'

curl -X POST http://localhost:3030/api/todos/add 
  -H "Content-Type: application/json" 
  -H "X-API-Key: your-api-key" 
  -d '{
    "user_id": "user-1",
    "content": "Fix authentication bug",
    "project": "Backend",
    "priority": "high",
    "contexts": ["@computer"]
  }'

Performance

Operation	Latency
Store memory	55-60ms
Semantic search	34-58ms
Tag search	~1ms
Entity lookup	763ns
Graph traversal (3-hop)	30µs

Compared to Alternatives

	Shodh-Memory	Mem0	Cognee
Deployment	Single 17MB binary	Cloud API	Neo4j + Vector DB
Offline	100%	No	Partial
Learning	Hebbian + decay + LTP	Vector similarity	Knowledge graphs
Latency	Sub-millisecond	Network-bound	Database-bound

Platform Support

Platform	Status
Linux x86_64	Supported
Linux ARM64	Supported
macOS ARM64 (Apple Silicon)	Supported
macOS x86_64 (Intel)	Supported
Windows x86_64	Supported

Production Deployment

Shodh-Memory is designed for single-machine deployments where multiple AI agents share a common memory store. For production use:

Security Model

Internet → Reverse Proxy (TLS + Auth) → Shodh-Memory (localhost:3030)

TLS/HTTPS: The server does not handle TLS directly. For network deployments, place it behind a reverse proxy (Nginx, Caddy, Traefik, Cloudflare Tunnel) that handles TLS termination.

Authentication: All data endpoints require API key authentication via X-API-Key header. Health and metrics endpoints are public for monitoring.

Network Binding: By default, the server binds to 127.0.0.1 (localhost only). Set SHODH_HOST=0.0.0.0 only when behind an authenticated reverse proxy.

Environment Variables

# Required for production
SHODH_ENV=production              # Enables production mode (stricter validation)
SHODH_API_KEYS=key1,key2,key3     # Comma-separated API keys

# Optional
SHODH_HOST=127.0.0.1              # Bind address (default: localhost)
SHODH_PORT=3030                   # Port (default: 3030)
SHODH_MEMORY_PATH=/var/lib/shodh  # Data directory
SHODH_REQUEST_TIMEOUT=60          # Request timeout in seconds
SHODH_MAX_CONCURRENT=200          # Max concurrent requests
SHODH_CORS_ORIGINS=https://app.example.com  # Allowed CORS origins

Example: Nginx Reverse Proxy

server {
    listen 443 ssl;
    server_name memory.example.com;

    ssl_certificate /etc/letsencrypt/live/memory.example.com/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/memory.example.com/privkey.pem;

    location / {
        proxy_pass http://127.0.0.1:3030;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Example: Caddy (Auto-TLS)

memory.example.com {
    reverse_proxy localhost:3030
}

Docker Compose (Production)

version: '3.8'
services:
  shodh-memory:
    image: roshera/shodh-memory:latest
    environment:
      - SHODH_ENV=production
      - SHODH_HOST=0.0.0.0
      - SHODH_API_KEYS=${SHODH_API_KEYS}
    volumes:
      - shodh-data:/data
    networks:
      - internal

  caddy:
    image: caddy:latest
    ports:
      - "443:443"
    volumes:
      - ./Caddyfile:/etc/caddy/Caddyfile
    networks:
      - internal

volumes:
  shodh-data:

networks:
  internal:

Community Implementations

Project	Description	Author
SHODH on Cloudflare	Edge-native implementation on Cloudflare Workers with D1, Vectorize, and Workers AI	@doobidoo

Have an implementation? Open a discussion to get it listed.

References

[1] Cowan, N. (2010). The Magical Mystery Four: How is Working Memory Capacity Limited, and Why? Current Directions in Psychological Science.

[2] Magee, J.C., & Grienberger, C. (2020). Synaptic Plasticity Forms and Functions. Annual Review of Neuroscience.

[3] Subramanya, S.J., et al. (2019). DiskANN: Fast Accurate Billion-point Nearest Neighbor Search. NeurIPS 2019.

License

Apache 2.0

---

MCP Registry · PyPI · npm · crates.io · Docs