JAEGIS RAVERSE MCP Server

Production-ready Model Context Protocol (MCP) server for RAVERSE - AI Multi-Agent Binary Patching System.

📊 Package Distribution

Overview

This MCP server exposes 35 core capabilities from RAVERSE 2.0 as standardized MCP tools, enabling seamless integration with Claude, other AI models, and external systems.

Features

Binary Analysis Tools (4 tools)

• Disassemble Binary: Convert machine code to human-readable assembly
• Generate Code Embedding: Create semantic vectors for code snippets
• Apply Patch: Programmatically modify binary files
• Verify Patch: Confirm patch application and integrity

Knowledge Base & RAG Tools (4 tools)

• Ingest Content: Add content to knowledge base
• Search Knowledge Base: Find relevant content via semantic search
• Retrieve Entry: Get specific knowledge base entries
• Delete Entry: Remove entries from knowledge base

Web Analysis Tools (5 tools)

• Reconnaissance: Gather intelligence about web targets
• Analyze JavaScript: Extract logic and API calls from JS code
• Reverse Engineer API: Generate OpenAPI specs from traffic
• Analyze WASM: Decompile and analyze WebAssembly modules
• Security Analysis: Identify vulnerabilities and security issues

Infrastructure Tools (5 tools)

• Database Query: Execute parameterized database queries
• Cache Operation: Manage Redis cache operations
• Publish Message: Send A2A protocol messages
• Fetch Content: Download web content with retry logic
• Record Metric: Track performance metrics

Advanced Analysis Tools (5 tools)

• Logic Identification: Identify logic patterns in code
• Traffic Interception: Intercept and analyze network traffic
• Generate Report: Generate comprehensive analysis reports
• RAG Orchestration: Execute RAG workflow
• Deep Research: Perform deep research on topics

Management Tools (4 tools)

• Version Management: Manage component versions
• Quality Gate: Enforce quality standards
• Governance Check: Check governance rules
• Generate Document: Generate structured documents

Utility Tools (5 tools)

• URL Frontier Operation: Manage URL frontier for crawling
• API Pattern Matcher: Identify API patterns in traffic
• Response Classifier: Classify HTTP responses
• WebSocket Analyzer: Analyze WebSocket communication
• Crawl Scheduler: Schedule crawl jobs

System Tools (4 tools)

• Metrics Collector: Record performance metrics
• Multi-Level Cache: Manage multi-level cache
• Configuration Service: Access configuration
• LLM Interface: Interface with LLM provider

NLP & Validation Tools (2 tools)

• Natural Language Interface: Process natural language commands
• PoC Validation: Validate vulnerabilities with PoC

Total: 35 Tools Across 9 Categories

Installation

Prerequisites

• Python 3.13+
• PostgreSQL 17 with pgvector
• Redis 8.2
• OpenRouter API key (for LLM features)

Quick Start (Recommended)

Option 1: NPX (Fastest - No Installation Required)

# Run the latest version without installation
npx -y raverse-mcp-server@latest

# Or with specific version
npx -y [email protected]

# Verify it works
npx -y raverse-mcp-server@latest -- --version

Option 2: NPM (Global Installation)

# Install globally
npm install -g raverse-mcp-server

# Run the server
raverse-mcp-server

# Verify installation
raverse-mcp-server --version

Option 3: PyPI (Python Package)

# Install via pip
pip install jaegis-raverse-mcp-server

# Run the server
python -m jaegis_raverse_mcp_server.server

# Verify installation
python -m jaegis_raverse_mcp_server.server --version

Option 4: Docker

# Pull and run Docker image
docker run -d \
  -e DATABASE_URL="postgresql://user:pass@host/db" \
  -e REDIS_URL="redis://localhost:6379" \
  -e OPENROUTER_API_KEY="sk-or-v1-..." \
  -p 8000:8000 \
  raverse/mcp-server:latest

Detailed Setup (From Source)

1. Clone and navigate to directory:

cd jaegis-RAVERSE-mcp-server

2. Create virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -e .

4. Configure environment:

cp .env.example .env
# Edit .env with your configuration

Installation Guides

• Complete Installation Guide - Detailed instructions for all methods
• Quick Start Guide - Get running in 5 minutes
• MCP Client Setup - Configure 20+ MCP clients

Configuration

All configuration is managed via environment variables. See .env.example for all available options.

Key Settings

| Variable | Default | Description | |----------|---------|-------------| | LOG_LEVEL | INFO | Logging level (DEBUG, INFO, WARNING, ERROR) | | DATABASE_URL | localhost | PostgreSQL connection string | | REDIS_URL | localhost | Redis connection string | | LLM_API_KEY | - | OpenRouter API key | | ENABLE_* | true | Feature flags for tool categories |

Client Configuration Examples

1. Standard Configuration (Claude Desktop, Cursor, Cline)

Add this to your MCP settings file (e.g., claude_desktop_config.json):

{
  "mcpServers": {
    "raverse": {
      "command": "npx",
      "args": ["-y", "raverse-mcp-server@latest"],
      "env": {
        "PROXY_URL": "https://raverse-mcp-proxy.use-manus-ai.workers.dev",
        "BACKEND_URL": "https://jaegis-raverse.onrender.com",
        "DATABASE_URL": "postgres://avnadmin:AVNS_8DLEfx78cGBg-xL6w_r@raverse-pg-db-raverse-pg-db.i.aivencloud.com:23055/defaultdb?sslmode=require",
        "REDIS_URL": "rediss://default:AVNS_R6qi29fLEGC3FGGW3Pg@raverse-valkey-cache-raverse-pg-db.g.aivencloud.com:23056",
        "LOG_LEVEL": "INFO",
        "SERVER_VERSION": "1.0.14"
      }
    }
  }
}

2. Advanced Configuration (With Proxy Disabled)

Use this if you want to run locally without the proxy:

{
  "mcpServers": {
    "raverse": {
      "command": "npx",
      "args": ["-y", "raverse-mcp-server@latest"],
      "env": {
        "SERVER_VERSION": "1.0.14",
        "DATABASE_URL": "postgres://avnadmin:AVNS_8DLEfx78cGBg-xL6w_r@raverse-pg-db-raverse-pg-db.i.aivencloud.com:23055/defaultdb?sslmode=require",
        "REDIS_URL": "rediss://default:AVNS_R6qi29fLEGC3FGGW3Pg@raverse-valkey-cache-raverse-pg-db.g.aivencloud.com:23056",
        "LOG_LEVEL": "ERROR"
      },
      "disabled": false
    },
    "raverse-mcp-proxy": {
      "command": "node",
      "args": ["-e", "process.stderr.write('proxy disabled\\n')"],
      "disabled": true
    }
  }
}

raverse-mcp-server

Programmatic Usage

from jaegis_raverse_mcp_server import MCPServer

server = MCPServer()
result = await server.handle_tool_call(
    "disassemble_binary",
    {"binary_path": "/path/to/binary"}
)

Tool Reference

Binary Analysis

disassemble_binary

Disassemble a binary file and extract structural information.

Parameters:

• binary_path (str): Path to binary file
• architecture (str, optional): Target architecture

Returns:

• binary_hash: SHA256 hash
• file_size: Size in bytes
• status: Operation status

generate_code_embedding

Generate semantic embedding for code content.

Parameters:

• code_content (str): Code to embed
• model (str): Embedding model name

Returns:

• content_hash: Content hash
• status: Operation status

apply_patch

Apply patches to binary file.

Parameters:

• binary_path (str): Binary file path
• patches (list): List of patch objects
• backup (bool): Create backup before patching

Returns:

• patch_count: Number of patches applied
• status: Operation status

verify_patch

Verify patch was applied correctly.

Parameters:

• original_binary (str): Original binary path
• patched_binary (str): Patched binary path

Returns:

• original_hash: Original binary hash
• patched_hash: Patched binary hash
• hashes_match: Whether hashes match

Knowledge Base

ingest_content

Add content to knowledge base.

Parameters:

• content (str): Content to ingest
• metadata (dict, optional): Associated metadata

Returns:

• content_hash: Content hash
• status: Operation status

search_knowledge_base

Search for relevant content.

Parameters:

• query (str): Search query
• limit (int): Max results (1-100)
• threshold (float): Similarity threshold (0-1)

Returns:

• query: Original query
• status: Operation status

retrieve_entry

Get specific knowledge base entry.

Parameters:

• entry_id (str): Entry ID

Returns:

• entry_id: Entry ID
• status: Operation status

delete_entry

Delete knowledge base entry.

Parameters:

• entry_id (str): Entry ID

Returns:

• entry_id: Entry ID
• status: Operation status

Web Analysis

reconnaissance

Perform web reconnaissance on target.

Parameters:

• target_url (str): Target URL

Returns:

• target_url: Target URL
• status: Operation status

analyze_javascript

Analyze JavaScript code.

Parameters:

• js_code (str): JavaScript code
• deobfuscate (bool): Deobfuscate code

Returns:

• status: Operation status
• endpoints_found: Number of endpoints found

reverse_engineer_api

Reverse engineer API from traffic.

Parameters:

• traffic_data (dict): Traffic data
• js_analysis (dict, optional): JS analysis results

Returns:

• status: Operation status

analyze_wasm

Analyze WebAssembly module.

Parameters:

• wasm_data (bytes): WASM module data

Returns:

• status: Operation status
• wasm_size: Module size

security_analysis

Perform security analysis.

Parameters:

• analysis_data (dict): Analysis data
• check_headers (bool): Check security headers
• check_cves (bool): Check for CVEs

Returns:

• status: Operation status

Infrastructure

database_query

Execute database query.

Parameters:

• query (str): SQL query
• params (list, optional): Query parameters

Returns:

• status: Operation status

cache_operation

Perform cache operation.

Parameters:

• operation (str): Operation (get, set, delete, exists, clear)
• key (str): Cache key
• value (any, optional): Value for set operation
• ttl (int, optional): Time to live in seconds

Returns:

• status: Operation status

publish_message

Publish A2A message.

Parameters:

• channel (str): Channel name
• message (dict): Message content

Returns:

• status: Operation status

fetch_content

Fetch content from URL.

Parameters:

• url (str): URL to fetch
• timeout (int): Timeout in seconds
• retries (int): Number of retries

Returns:

• status: Operation status

record_metric

Record performance metric.

Parameters:

• metric_name (str): Metric name
• value (float): Metric value
• labels (dict, optional): Metric labels

Returns:

• status: Operation status

Error Handling

All tools return structured error responses:

{
  "success": false,
  "error": "Error message",
  "error_code": "ERROR_TYPE"
}

Error Codes

• VALIDATION_ERROR: Input validation failed
• DATABASE_ERROR: Database operation failed
• CACHE_ERROR: Cache operation failed
• BINARY_ANALYSIS_ERROR: Binary analysis failed
• WEB_ANALYSIS_ERROR: Web analysis failed
• TOOL_EXECUTION_ERROR: Tool execution failed
• UNKNOWN_TOOL: Tool not found

Logging

Structured logging with JSON output. Configure via LOG_LEVEL environment variable.

Performance

• Concurrent Tasks: Configurable via MAX_CONCURRENT_TASKS
• Cache TTL: Configurable via CACHE_TTL_SECONDS
• Request Timeout: Configurable via REQUEST_TIMEOUT_SECONDS

Security

• Input validation on all parameters
• Parameterized database queries
• Secure credential management via environment variables
• No hardcoded secrets

Distribution

RAVERSE MCP Server is available through multiple distribution channels:

NPM Package

npm install -g @raverse/mcp-server

• Package: @raverse/mcp-server
• Registry: https://registry.npmjs.org/

PyPI Package

pip install jaegis-raverse-mcp-server

• Package: jaegis-raverse-mcp-server
• Registry: https://pypi.org/

Docker Image

docker pull raverse/mcp-server:latest

• Registry: Docker Hub
• Image: raverse/mcp-server

MCP Client Integration

The RAVERSE MCP Server is compatible with 20+ AI coding assistants and IDEs:

Anthropic

• Claude Desktop - macOS, Windows, Linux
• Claude Web - Web-based interface

Code Editors

• Cursor IDE - AI-first code editor
• VS Code + Cline - VSCode extension
• VS Code + Roo Code - VSCode extension
• Windsurf IDE - Codeium's IDE
• Zed Editor - High-performance editor

Web-Based IDEs

• Replit - Cloud IDE
• Bolt.new - AI web development
• v0.dev - Vercel's AI builder
• Lovable.dev - AI web builder

AI Coding Assistants

• Augment Code - Code intelligence
• Manus AI - AI agent
• Devin AI - Autonomous AI engineer
• Continue.dev - Open-source copilot
• Aider - AI pair programmer

Enterprise & Cloud

• JetBrains AI Assistant - IntelliJ, PyCharm, etc.
• GitHub Copilot - GitHub's AI assistant
• Sourcegraph Cody - Code search AI
• Tabnine - AI code completion
• Amazon CodeWhisperer - AWS AI assistant

Research & Analysis

• Perplexity - AI research assistant
• GPT-4 Web - OpenAI's web interface

Configuration Files: See /mcp-configs/ directory for all client configurations

Setup Guide: See MCP_SETUP_GUIDE.md for detailed instructions for each client

Troubleshooting

Issue: Server shows wrong version or no tools in Augment Code

Symptoms:

• Augment Code shows red dot with no tool count
• Server reports version 1.0.0 instead of 1.0.10
• Redis/PostgreSQL connection errors on startup

Solution:

1. Clear caches:

   npm cache clean --force
   pip cache purge

2. Update Augment Code MCP configuration to use v1.0.10:

   {
     "mcpServers": {
       "raverse": {
         "command": "npx",
         "args": ["-y", "[email protected]"]
       }
     }
   }

3. Restart Augment Code completely (close and reopen)

4. Wait 15-20 seconds for tool discovery

Expected Result:

raverse (35) tools ✅ (green indicator)

Issue: Python not found

Solution:

# Install Python 3.13+
# Windows: https://www.python.org/downloads/
# macOS: brew install [email protected]
# Linux: apt-get install python3.13

Issue: Redis/PostgreSQL connection errors

Solution: The server uses lazy initialization - database connections only happen on first tool request. If you see connection errors:

1. Ensure PostgreSQL and Redis are running
2. Check DATABASE_URL and REDIS_URL in .env
3. Run setup wizard: raverse-mcp-server --setup

Issue: Port already in use

Solution:

# Change port in .env
SERVER_PORT=8001

Infrastructure & Hosting

Free Hosting Architecture

For organizations looking to deploy RAVERSE MCP server with minimal infrastructure costs, we provide a comprehensive guide on setting up a Hybrid-Cloud Architecture for free hosting.

📄 Free Hosting Setup Using a Hybrid-Cloud Architecture

This guide covers:

• Cost-effective deployment strategies
• Hybrid-cloud architecture design
• Free tier services integration
• Scalability and performance optimization
• Security best practices for free hosting

Perfect for startups, research projects, and proof-of-concept deployments.

Documentation

• Installation Guide - Complete installation instructions
• Quick Start - Get running in 5 minutes
• MCP Setup Guide - Setup for 20+ MCP clients
• MCP Client Setup - Configure MCP clients
• Integration Guide - Integrate with RAVERSE
• Deployment Guide - Production deployment
• Tools Registry - Complete tool reference (all 35 tools)
• Package Distribution - For package maintainers
• Publishing Guide - Publishing to npm and PyPI

License

MIT License - See LICENSE file for details

Support

For issues and questions, please refer to the main RAVERSE repository.