Introduction

Coding with AI agents sometimes can be boring, not because they lack intelligence, but because they lack memory.

Every new session feels like déjà vu:

• you re-explain the same requirements
• re-share the same code
• re-trace steps the agent has already helped you with.

The moment you close your IDE, all that context is gone. And when teammates use their own agents, there’s no continuity, everyone starts from scratch.

This is where Regolo.ai + Cipher changes the game. Cipher adds persistent memory and reasoning layers, transforming short-term helpers into long-term coding partners. Knowledge survives across sessions, agents understand your codebase as it evolves, and context can be shared seamlessly within your team.

With Regolo.ai as the engine behind it, you get powerful models guided by Cipher’s advanced memory orchestration.

Key Features

• 🔌 MCP integration across IDEs
• 🧠 Persistent AI memory
• 🔄 Switch IDEs without losing context
• 👥 Shared team memory
• ⚙️ Install on your IDE with zero configuration needed.
• 🧬 Dual Memory Layer that captures
  • System 1 (Programming Concepts & Business Logic & Past Interaction)
  • System 2 (reasoning steps of the model when generating code).
    
    
    

Prerequisites

Before you begin, make sure you have:

1. Node.js (v18 or later)
2. A Regolo.ai API Key – available at regolo.ai
3. An MCP-enabled IDE (VS Code, Cursor, Claude Desktop, etc.)

Installing Cipher

NPM Package (Recommended for Most Users)

# Install globally
npm install -g @byterover/cipher

# Or install locally in your project
npm install -g @byterover/cipherCode language: PHP (php)

Docker

# Clone and setup
git clone https://github.com/campfirein/cipher.git
cd cipher

# Configure environment
cp .env.example .env
# Edit .env with your API keys

# Start with Docker
docker-compose up --build -d

# Test
curl http://localhost:3000/healthCode language: PHP (php)

From Source

pnpm i && pnpm run build && npm link

CLI Usage 💻

# Interactive mode
cipher

# One-shot command
cipher "Add this to memory as common causes of 'CORS error' in local dev with Vite + Express."

# API server mode
cipher --mode api

# MCP server mode
cipher --mode mcp

# Web UI mode
cipher --mode uiCode language: PHP (php)

The Cipher Web UI provides an intuitive interface for interacting with memory-powered AI agents, featuring session management, tool integration, and real-time chat capabilities.

Using Regolo

Cipher supports multiple configuration options for different deployment scenarios. The main configuration file is located at memAgent/cipher.yml.

1. Example YAML

This is an example of .yml file for the Regolo integration:

# List of MCP servers to use
# mcpServers: 
#   filesystem:
#     type: stdio
#     command: npx
#     args:
#       - -y
#       - '@modelcontextprotocol/server-filesystem'
#       - .

# If not specified, please set it like this:
mcpServers: {}

# Choose ONLY ONE of the following LLM providers
llm:
  provider: openai
  model: Llama-3.3-70B-Instruct
  apiKey: $REGOLO_API_KEY
  baseURL: https://api.regolo.ai/v1
  maxIterations: 50
  
# Embedding configuration: ONLY ONE of the following
# OpenAI:
embedding:
  provider: openai
  model: gte-Qwen2
  apiKey: $REGOLO_API_KEY
  baseURL: https://api.regolo.ai/v1

# Disable embeddings entirely:
# embedding:
#   disabled: true



# Evaluation LLM configuration (non-thinking model for evaluation step)
evalLlm:
  provider: openai
  model: Llama-3.1-8B-Instruct
  apiKey: $REGOLO_API_KEY
  baseURL: https://api.regolo.ai/v1


# System prompt - User customizable
# This prompt will be combined with built-in tool usage instructions
systemPrompt:
  enabled: true
  content: |
    You are an AI programming assistant focused on coding and reasoning tasks. You excel at:
    - Writing clean, efficient code
    - Debugging and problem-solving
    - Code review and optimization
    - Explaining complex technical concepts
    - Reasoning through programming challenges
Code language: YAML (yaml)

2. Environment Variables 🔐

Create a .env file in your project root with these essential variables:

cp .env.example .envCode language: CSS (css)

# Cipher Environment Configuration
# Copy this file to .env and fill in your actual values

# ====================
# API Configuration
# ====================
# At least one API key is required
REGOLO_API_KEY=<your regolo api key>

# ====================
# Application Settings
# ====================
NODE_ENV=development
CIPHER_LOG_LEVEL=info
REDACT_SECRETS=true

# ====================
# Storage Configuration
# ====================
STORAGE_CACHE_TYPE=in-memory
# STORAGE_CACHE_HOST=localhost
# STORAGE_CACHE_PORT=6379
# STORAGE_CACHE_USERNAME=
# STORAGE_CACHE_PASSWORD=
# STORAGE_CACHE_DATABASE=0

STORAGE_DATABASE_TYPE=in-memory
STORAGE_DATABASE_NAME=cipher.db
STORAGE_DATABASE_PATH=./data

CIPHER_MULTI_BACKEND=1
CIPHER_PG_URL=postgres://cipheruser:cipherpass@localhost:5432/cipherdb
CIPHER_WAL_FLUSH_INTERVAL=5000

# ====================   
# Memory Search Configuration
# ====================
# Options: knowledge, reflection, both
SEARCH_MEMORY_TYPE=both

# ====================
# Vector Store Configuration
# ====================
# Available types: in-memory, qdrant, milvus
VECTOR_STORE_TYPE=in-memory

# Qdrant configuration (only used if VECTOR_STORE_TYPE=qdrant)
# VECTOR_STORE_HOST=localhost
# VECTOR_STORE_PORT=6333
# VECTOR_STORE_URL=http://localhost:6333
# VECTOR_STORE_API_KEY=your-qdrant-api-key

# Milvus configuration (only used if VECTOR_STORE_TYPE=milvus)
# VECTOR_STORE_HOST=localhost
# VECTOR_STORE_PORT=19530
# VECTOR_STORE_URL=your-milvus-cluster-endpoint
# VECTOR_STORE_API_KEY=your-milvus-api-key
# VECTOR_STORE_USERNAME=your-milvus-zilliz-cloud-username
# VECTOR_STORE_PASSWORD=your-milvus-zilliz-cloud-password

# Vector collection settings
VECTOR_STORE_COLLECTION=knowledge_memory
# VECTOR_STORE_COLLECTION_NAME=your_collection_name
VECTOR_STORE_DIMENSION=1536
VECTOR_STORE_DISTANCE=Cosine
VECTOR_STORE_ON_DISK=false
# VECTOR_STORE_MAX_VECTORS=10000

# ====================
# Reflection Memory Configuration
# ====================
REFLECTION_VECTOR_STORE_COLLECTION=reflection_memory
DISABLE_REFLECTION_MEMORY=true

# Enable or disable input refinement for better retrieval
ENABLE_QUERY_REFINEMENT=true Code language: Python (python)

3. MCP Server Usage with RooCode

Now let’s use Cipher with RooCode, if you lost it, there is the guide to use Regolo models with RooCode: https://regolo.ai/roo-code-ai-powered-autonomous-coding-in-vscode/

Example Video

Step-to-Step Guide

1. Open RooCode and click on the three dots on the right-top corner:
2. Click on “MCP Servers”
3. Click on Edit Global MCP
4. Paste this code:
   

{
  "mcpServers": {
    "cipher": {
      "type": "stdio",
      "command": "bash",
      "args": [
        "-lc",
        "source <PATH to your venv> && cd <PATH to your cipher folder> && cipher --mode mcp"
      ],
      "env": {
        "REGOLO_API_KEY": "<your-api-key>",
        "MCP_SERVER_MODE": "default"
      },
      "disabled": false,
      "alwaysAllow": []
    }
  }
}Code language: JSON / JSON with Comments (json)

You should see:

Use Case