AI-Native Web Traffic Debugging Tool

English | 简体中文

RelayCraft uses a modular 3-layer architecture to provide reliable traffic debugging with a modern interface.

graph LR
    %% Style definitions
    classDef client fill:#f9f9f9,stroke:#333,stroke-width:2px,color:#333
    classDef engine fill:#e3f2fd,stroke:#1565c0,stroke-width:2px,color:#000
    classDef backend fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000
    classDef frontend fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px,color:#000
    classDef ai fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px,color:#000,stroke-dasharray: 5 5

    %% Node definitions
    Traffic(["🌐 Web Traffic<br/>(HTTP/HTTPS/WS)"]):::client
    
    subgraph Core ["RelayCraft Architecture"]
        direction LR
        Engine["<b>Engine Layer</b><br/>Python / mitmproxy"]:::engine
        Backend["<b>System Layer</b><br/>Tauri / Rust"]:::backend
        UI["<b>UI Layer</b><br/>React / Zustand"]:::frontend
    end
    
    MCP["<b>🤖 AI Agent</b><br/>Claude, Cursor, etc."]:::ai

    %% Connections
    Traffic <==>|Proxy/Intercept| Engine
    Engine <==>|IPC / HTTP API| Backend
    Backend <==>|Tauri Commands| UI
    
    MCP <.->|Model Context Protocol| Backend

RelayCraft is an AI-native network debugging tool for modern development. Built with Tauri, React, and Rust, it combines a proxy engine, AI-assisted workflows, and an extensible plugin system — all running locally, with zero accounts required.

• 🤖 AI-Native Workflows: Create rewrite rules in natural language, analyze failed requests, and search traffic with plain language in one interface.
• 🔌 MCP Server: Expose live traffic data and rule management to any external AI tool (Claude Desktop, Cursor, etc.) via the Model Context Protocol. Let your AI agent debug alongside you.
• 🏗️ Modern Architecture: A lightweight core built with Tauri and Rust, powered by the mitmproxy engine.
• 🛡️ Privacy First: Your data stays yours. Fully offline, zero accounts, local storage, local AI support, open source, and no telemetry.
• 🐍 Scriptable & Extensible: Build custom traffic logic with mitmproxy Python hooks and extend the workspace through plugins and theming.

• 📡 Capture: Inspect HTTP, HTTPS, and WebSocket traffic in real-time.
• 🔍 Filter: Filter by method, domain, status code, or content type using a powerful query syntax.
• 💎 Detail Inspection: JSON highlighting, image previews, and structured header/body views.
• 📦 Export: One-click export to cURL, HAR, or Relay Session (.relay).

Manage traffic behavior with a visual rule builder — no config files needed. Supports 6 rule types:

Real-time control to pause, edit, and resume flows in both Request and Response phases.

• Smart Matching: Support for wildcard path matching and RegEx for precise interception.
• On-the-fly Editing: Modify headers and body content before they reach the server or client.
• Edge Case Simulation: Useful for testing error handling, mock responses, and race-condition debugging.

A built-in API client designed for debugging, deeply integrated with your traffic history.

• Direct Replay: Swiftly replay any captured flow with a dedicated visual editor.
• Format Support: Full management of headers and body content (Raw, Form-Data, JSON).
• cURL Integration: Import requests from cURL strings with a single click.
• Instant Preview: View formatted responses (JSON, HTML, Assets) directly in the composer panel.

Use the mitmproxy Python ecosystem for scenarios beyond visual rules.

• Native Hooks: Write custom logic using standard mitmproxy event hooks.
• Modern Editor: Powered by CodeMirror 6 with syntax highlighting and auto-formatting.
• Dedicated Logging: Script outputs are centralized in a dedicated tab within the system logs for easy troubleshooting.

Global Ctrl(⌘) + K command center with context-aware AI across every workflow:

• Natural Language Rules: Describe what you want intercepted or rewritten; AI builds the rule.
• Request Analysis: Generate request summaries and diagnostics, including rule hits, potential security issues, and optimization hints.
• Smart Search: Find specific traffic using natural language queries.
• Script Generation: Generate Python mitmproxy scripts from plain descriptions.

RelayCraft runs a built-in MCP (Model Context Protocol) server, letting any compatible AI client connect and work with your live traffic data.

Read tools (no auth required — zero config):

• list_sessions / list_flows / get_flow / search_flows / get_session_stats / list_rules

Write tools (Bearer token, shown in Settings → Integrations):

• create_rule — create any of the 6 rule types using natural language parameters
• delete_rule / toggle_rule — manage rules in the session
• replay_request — replay captured traffic through the proxy

Compatible with Claude Desktop, Cursor, Windsurf, and any tool supporting the MCP HTTP transport.

• Node.js 20.19+ or 22.12+ (Vite compatibility; Node 21 is not supported)
• pnpm 9+ (install; this repo uses lockfile v9)
• Rust (stable toolchain)
• Python (3.10+)

# 1. Clone the repository
git clone https://github.com/relaycraft/relaycraft.git
cd relaycraft

# 2. Set up the Python Engine
# Follow the instructions in engine-core/README.md
# to build and place the engine binary in the required directory.

# 3. Install frontend dependencies
pnpm install

# 4. Start development mode
pnpm tauri dev

Pre-built binaries for macOS, Windows, and Linux are available on the Releases Page.

• Contributing Guide — Learn how to contribute to the project.
• Roadmap — Product direction and upcoming milestones (GitHub Discussion).
• Commercial Use — Sponsorship and enterprise licensing information.
• Plugin Registry — Explore community-built plugins.

RelayCraft is open-sourced under the GNU GPL v3 or later. See the LICENSE file for details.

Crafted with ❤️ by the RelayCraft Team.