mcp-agent-tools/README.md

🤖 MCP Agent Tools & Multimodal Support

A FastAPI-based AI agent tool service providing agent core tools, multimodal AI capabilities, TTS/OCR, knowledge base, essays, and utility APIs.

This project contains two FastAPI services for AI agent assistance:

1. Main Agent Tools — core agent tools: MCP endpoint, safe command execution, workspace switching, SSH session management, and permission dialogs.
2. Multimodal & Gadget Support — multimodal AI (Qwen chat/ask/TTS/OCR), knowledge base management, essay recording/search, and utility APIs (WHOIS / IP / GitHub).

---

📦 Components

🤖 Main Agent Tools (agent_tools.py)

Agent core tools exposed via FastAPI:

• MCP endpoint (/mcp)
• Safe command execution (execute_command_safe)
• Workspace switching (/switch_workspace, /current_workspace)
• SSH session management (/ssh/sessions, /ssh/cleanup)
• Python runner (create_and_run_python)
• Permission dialogs (confirm / input) on Windows

🧩 Multimodal & Gadget Support (multimodal_gadget_support.py)

• AI Chat / Ask — Qwen (Flash / Pro) chat & ask endpoints
• TTS — text-to-speech (Flash / Plus / Max)
• OCR — text recognition (/ocr, /ocr_advanced)
• Knowledge Base — add / query / search / manage knowledge points
• Essays — record, search, view personal essays
• Utilities — WHOIS, IP info, GitHub repo lookup
• Keyword split — DeepSeek-based query segmentation

---

🚀 Quick Start

Prerequisites

• Python 3.9+
• fastapi, uvicorn, pydantic
• Optional: paramiko (SSH), openai (AI), pandas, etc.

Run

pip install fastapi uvicorn pydantic

# Run Agent Tools
python agent_tools.py

# Run Multimodal & Gadget Support (on another port)
python multimodal_gadget_support.py

---

📁 Project Structure

mcp/
├── Main Agent Tools.py              # 🤖 Agent core tools (FastAPI)
├── Multimodal and Gadget Support.py # 🧩 Multimodal & gadget support (FastAPI)
└── README.md                        # This document

---

📄 License

This project is licensed under the MIT License. See LICENSE for details.

---

⚠️ Security Note

This service executes system commands and exposes agent tools. Run only in trusted environments. All sensitive keys are configured via environment variables, not hardcoded.