websockets
Build correct, fast WebSocket servers and clients in Python with asyncio, threading, or Sans-I/O APIs.
Repository Health
Technical Analysis
websockets is a Python library for building WebSocket servers and clients with a focus on correctness, simplicity, robustness, and performance. Its default implementation is built on asyncio and exposes an elegant coroutine-based API, while additional threading and Sans-I/O implementations let you integrate WebSockets into any concurrency model.
Heavily tested for compliance with RFC 6455 and RFC 7692, websockets enforces 100% branch coverage in CI and ships a C extension with prebuilt wheels for fast, memory-efficient production deployments. All you need to remember is await ws.recv() and await ws.send(msg) — the library manages connections, ping/pong keepalives, backpressure, and clean closes for you.
What You Get
- An asyncio implementation with a coroutine-based
serve/connectAPI and async iteration over incoming messages - A synchronous
threadingimplementation for scripts and codebases that do not use asyncio - A Sans-I/O core (
websockets.protocol) you can drive from any I/O framework - A C speedups extension with prebuilt wheels for Linux, macOS, and Windows across Python versions
- Built-in permessage-deflate compression, basic-auth helpers, proxy support, and a
websocketsCLI client
Common Use Cases
- Serving realtime updates to browser or mobile clients over a persistent WebSocket connection
- Writing a WebSocket client to consume a streaming API or exchange feed
- Building the WebSocket transport layer underneath higher-level frameworks such as uvicorn or Sanic
- Bridging services with low-latency bidirectional messaging without polling
Under The Hood
Architecture — websockets is organized around a Sans-I/O core: src/websockets/protocol.py together with client.py and server.py implement the RFC 6455 state machine as pure functions over bytes and events, entirely decoupled from any I/O. Concrete transports wrap this core — src/websockets/asyncio/ provides the asyncio Connection, ServerConnection, and ClientConnection classes plus serve/connect, src/websockets/sync/ provides the threading equivalents, and src/websockets/legacy/ retains the previous asyncio implementation for compatibility. Framing, headers, URIs, and HTTP/1.1 handshake handling are factored into dedicated modules (frames.py, headers.py, uri.py, http11.py), and the public surface is re-exported lazily via imports.lazy_import in __init__.py. Tech Stack — Pure Python (requires Python >=3.11) built on the standard-library asyncio and threading modules with no runtime third-party dependencies; an optional C extension (speedups.c) accelerates frame masking and is shipped as prebuilt wheels via cibuildwheel. Packaging uses setuptools with configuration in pyproject.toml, and the project ships a py.typed marker for full type-checker support. Code Quality — Exceptionally strong: 59 test modules under tests/ mirror the source layout, CI fails under 100% branch coverage, and the codebase is fully type-annotated with from __future__ import annotations and precise typing throughout. Naming is consistent, docstrings are thorough, and error handling is modeled by a dedicated exceptions.py hierarchy (ConnectionClosed, InvalidHandshake, and friends). API Design — The public API is deliberately minimal and ergonomic: async with serve(handler, host, port) for servers, async with connect(uri) for clients, and just send/recv plus async iteration for messages. The same conceptual API is mirrored across the asyncio and threading implementations, documentation on Read the Docs is a first-class concern with a tutorial and reference, and connection management, backpressure, and keepalive are handled automatically so little boilerplate is required.
Used by 30 apps in this directory
Scalar
Developer Tools
Beautiful, interactive OpenAPI documentation with a built-in offline-first API client and multi-language code generation — all in one open-source platform.
Skyvern
AI Agents · Automation
Skyvern (YC S2023) automates browser-based workflows by pairing LLMs with computer vision, letting agents click, fill, and extract data on sites they've never seen, without brittle XPath selectors that break on every layout change.
VoiceStudio
Mcp · Music Audio · Voice AI
Open-source, fully local ElevenLabs alternative for voice cloning, voice design, video dubbing, dictation, transcription and audiobooks, with a local API and MCP server for agents.