See every request your apps make. Decrypt HTTPS in one click. Let your AI read it too.
A free, local HTTP and HTTPS debugging proxy with a fast desktop UI.
No account, no cloud, no paid tier.
Features · Install · AI assistants · Docs · Roadmap
| One click interception | Launch Chrome, Edge, Brave, Chromium or Vivaldi in an isolated window that is already routed through TrafficKit, HTTPS included. Or open a terminal where curl, Node, Python and git go through it automatically. |
| Real HTTPS decryption | A certificate authority generated on your machine. It is never added to your system trust store; only the windows TrafficKit opens trust it. |
| Live inspector | Headers, query, cookies, bodies and timings as they happen. Bodies are decoded (gzip, brotli, zstd) and JSON is formatted with line numbers. |
| WebSocket inspector | Every message in both directions, decoded live, including compressed frames. Filter by direction, search payloads, and read JSON messages formatted. |
| Built for big sessions | A virtualized table stays smooth with tens of thousands of requests. Filter with plain words, sort any column, move with the keyboard. |
| AI assistants over MCP | Claude, Cursor and any MCP client can list, search and inspect your traffic and open intercepted browsers. Passwords, cookies and tokens are redacted before anything leaves your machine. |
| Clear errors | DNS failures, refused connections, timeouts and certificate problems come with a plain explanation and a suggested fix. |
| Private by design | Everything stays in memory on your machine. No telemetry, no update pings, no sign in. |
TrafficKit is built from source for now. You need Go 1.25, Node 20 and, for the desktop app, Rust with the Tauri prerequisites.
git clone https://github.com/kspkr/TrafficKit.git
cd TrafficKit
npm install
# desktop app and installers
cd apps/desktop
npm run tauri:buildInstallers land in apps/desktop/src-tauri/target/release/bundle.
Just want to try it? Run the engine and UI in your browser:
npm run dev # UI on http://localhost:5173, proxy on 127.0.0.1:8877Or skip the UI entirely:
go run ./cmd/traffickit proxy
curl -x http://127.0.0.1:8877 http://example.com/TrafficKit ships a Model Context Protocol server. With the app running:
claude mcp add traffickit -- /path/to/traffickit mcpSettings → AI assistants shows the exact command for your install, plus the JSON config for Claude Desktop, Cursor and VS Code. Then just ask:
Open my app's login page in a TrafficKit browser. I'll sign in, you tell me why the request fails.
Which calls to api.stripe.com errored in the last few minutes, and what did they return?
The assistant can list, search and inspect traffic, read WebSocket messages, wait for a specific request, open and close browsers, and route its own commands through the proxy. The app shows when an assistant is connected. Tools, redaction and privacy details
All networking lives in a Go engine: the proxy, HTTPS interception, capture, the browser and terminal launchers, and the MCP server. It ships as a single binary that doubles as the command line tool.
The desktop app is a thin Tauri shell around a React UI. It starts the engine and talks to it over an authenticated local API, so the UI never touches a socket and the engine never touches pixels. The architecture notes go deeper.
| Getting started | Connecting browsers, terminals and devices |
| AI assistants | MCP setup, tools and redaction |
| Configuration | Config file, environment variables, flags |
| Security model | What TrafficKit protects and how |
| Development | Building, testing, releasing |
| Engine API | The local API the UI and MCP server use |
Issues and pull requests are welcome. Start with CONTRIBUTING.md, and report security problems privately as described in SECURITY.md.