The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
MCP Armor
MCP Armor continuously secures and monitors Model Context Protocol operations through static and dynamic scans, revealing hidden risks in agent-to-tool communications.
Install
pip install mcp-armorInstall command taken from the project's README or registry entry. Always review code before running it.
README excerpt
Read full README on GitHub →MCP Armor
🚀 Overview
MCP Armor is a comprehensive security scanner for Model Context Protocol (MCP). Automatically discovers, analyzes, and secures MCP servers integrated with all major Agentic IDEs, Agents and Clients.
💡 Features
- 🔍 Auto-Discovery: Finds known MCP configurations for popular Agentic IDEs like Cursor, Windsurf, VS Code, Claude Desktop, and more
- 🔧 Tool, Resource & Prompt Inventory: Connects to MCP servers and catalogs available tools, resources, and prompt templates
- 🛡️ Security Analysis: Specialized security checks including Prompt Injection, Rug Pull Attack, Cross-server Tool Shadowing, Tool Poisoning, Tool Name Ambiguity, and more..
- 🧭 Baseline Drift Detection: Captures approved MCP components and detects rug pulls attacks
- 📊 Comprehensive Reporting: Generates JSON and Markdown reports with actionable findings
- 📜 Audit Trail: Timestamped baselines and reports for full traceability of changes and findings
🧰 Installation
pip install mcp-armor
🏃 Quick Start
# Scan all configurations with security analysis (auto-detects baseline.json if present)
mcp-armor scan
# Inspect configurations and generate baseline (defaults to baseline.json)
mcp-armor inspect
# Use custom configuration file
mcp-armor scan --config /path/to/config.json
# Scan multiple configuration files
mcp-armor scan \
--config /path/to/cursor.mcp.json \
--config /path/to/vscode.mcp.json
# Use custom baseline file path
mcp-armor inspect --baseline /path/to/my-baseline.json
mcp-armor scan --baseline /path/to/my-baseline.json
# Generate markdown report
mcp-armor scan --report-type md
# Save to custom file
mcp-armor scan --output my-report.json
mcp-armor scan --report-type md --output my-report.md
⚙️ Command Options
| Option | Description |
|---|---|
--config |
Custom configuration file path (can be used multiple times) |
--baseline |
Baseline file for drift detection (scan) or creation (inspect) |
--report-type {json,md} |
Output format (default: json) |
--output |
Custom output file path |
--verbose |
Detailed terminal output |
--show-logs |
Display debug logs in terminal |
🔰 Security Checks
🛡️ Standard Checks
- Prompt Injection
- Indirect Prompt Injection
- Cross-Server Tool Shadowing
- Tool Poisoning
- Prompt Injection in Tool Description, Name and Args
- Command Injection in Tool Description, Name and Args
- Tool Name Ambiguity
- Command Injection
- Excessive Tool Permissions
- Hardcoded Secrets
🧭 Baseline Checks
Detects deviations from approved MCP components (requires a baseline generated via inspect mode):
- Rug Pull Attack
- Tool Modified
- Resource Modified
- Resource Template Modified
- Prompt Modified
📃 Logging
Logs are automatically saved to logs/mcp_armor.log:
# Default: logs saved to file only
mcp-armor scan
# Show logs in terminal too
mcp-armor scan --show-logs
🧪 Demo
Similar plugs
Reverse engineer anything with agents, from app behavior down to native binaries.
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype…
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)
A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io