Ffl

关于

ffl-mcp acts as a lightweight MCP server that integrates the power of 'ffl' (FastFileLink) with AI models like Claude and Codex. It allows AI to generate and manage secure, temporary P2P links to local resources without exposing their content to the LLM itself. This facilitates advanced collaborative workflows, such as AI-assisted debugging where models can share environment snapshots, logs, or database states with other instances or human developers, enabling efficient problem diagnosis and resolution through secure, direct peer-to-peer file transfer capabilities.

主要功能

• 1 GitHub stars
• Securely share text, base64 data, or files locally with end-to-end encryption by default
• Download files from P2P (WebRTC) or standard HTTP(S) links with resume capability
• Manage active sharing sessions, including listing, stopping, and retrieving details
• Avoids sending file contents to LLMs by triggering local file sharing commands
• Automatic configuration and installation for popular MCP clients like Claude Desktop/Code and Codex CLI

使用案例

• AI-driven collaborative debugging by sharing local environments, logs, and database states
• Securely sharing local files or sensitive data generated by AI models with other users or systems
• Allowing AI to download and process files from various sources, including P2P links