HTTPX
A next-generation HTTP client for Python with sync, async, and HTTP/2 support
Repository Health
Technical Analysis
HTTPX is a fully featured HTTP client for Python 3 that provides a broadly requests-compatible API while adding first-class support for both synchronous and asynchronous programming models. It supports HTTP/1.1 and HTTP/2 out of the box, ships an optional command-line client, and can dispatch requests directly to WSGI or ASGI applications for in-process testing.
Built by the Encode organization (also behind Starlette and Uvicorn), HTTPX layers a friendly, typed API on top of the lower-level httpcore transport library, giving developers connection pooling, streaming responses, proxy support, and strict timeout handling without sacrificing the ergonomics that made requests popular.
What You Get
- A
Client/AsyncClientpair with identical APIs for sync and async code, plus module-levelhttpx.get()/httpx.post()shortcuts - Native HTTP/2 support (opt-in via the
http2extra) alongside HTTP/1.1 - Direct-to-app transports (
WSGITransport,ASGITransport) for testing web apps without a running server - Connection pooling, streaming request/response bodies, and multipart file uploads
- A pluggable
httpcore-based transport layer with aMockTransportfor unit-testing HTTP calls - An optional CLI (
httpx[cli]) for making requests from the terminal with syntax-highlighted output
Common Use Cases
- Calling external REST APIs from a Python backend service, sync or async
- Writing integration tests for FastAPI/Starlette apps via
ASGITransportwithout binding a real port - Streaming large file downloads or uploads with bounded memory usage
- Migrating an existing
requests-based codebase to async with minimal API changes - Building HTTP/2-aware clients that need multiplexed connections to modern APIs
Under The Hood
Architecture — HTTPX separates concerns cleanly across a few layers: _client.py (2,019 lines) defines a BaseClient holding shared state (auth, headers, cookies, timeouts, redirect policy) that both the synchronous Client and asynchronous AsyncClient subclass; _api.py exposes stateless module-level functions (get, post, request, stream, etc.) that construct a short-lived Client under the hood for one-off calls; _models.py (1,277 lines) implements Request/Response/Headers/URL value objects; and the _transports/ package (default.py, asgi.py, wsgi.py, mock.py, base.py) implements the actual wire-level dispatch, delegating real network I/O to the httpcore library while ASGITransport/WSGITransport bypass the network entirely to call an application object in-process — this is what makes HTTPX popular for testing async web frameworks.
Tech Stack — Pure Python 3.9+, built with hatchling/hatch-fancy-pypi-readme. Core runtime dependencies are minimal and deliberate: httpcore==1.* (transport), certifi (CA bundle), idna (IDN support), and anyio (async concurrency abstraction, enabling both asyncio and Trio backends). Optional extras (h2 for HTTP/2, socksio for SOCKS proxies, click/rich/pygments for the CLI, brotli/zstandard for extra content-encoding support) keep the base install lean while allowing opt-in capability.
Code Quality — The project advertises and maintains 100% test coverage (tool.coverage.run config enforces this), with 37 test modules under tests/ covering client behavior, transports, models, and edge cases like redirects and multipart encoding. Strict mypy mode is enabled project-wide (tool.mypy: strict = true), and ruff lint rules (E, F, I, B, PIE) are enforced. Internal modules use a leading-underscore convention (_client.py, _models.py) to clearly separate public API surface (re-exported via __init__.py) from implementation detail.
API Design — HTTPX deliberately mirrors the requests API (httpx.get(url), response .json()/.text/.status_code) so migration is close to a drop-in replacement, while extending it with async def equivalents on AsyncClient using the identical method names and signatures. Context-manager usage (with httpx.Client() as client: / async with httpx.AsyncClient() as client:) is the documented idiom for connection reuse, and the library is fully type-annotated, giving strong editor autocomplete and static-analysis support out of the box.
Used by 86 apps in this directory
NoteDiscovery
Knowledge Management · Note Taking · Productivity
A self-hosted markdown knowledge base with MCP-powered AI integration, graph views, LaTeX math, and built-in drawing — your notes stay on your server, forever.
oh-my-claudecode
AI Agents
Multi-agent orchestration for Claude Code with zero learning curve — install as a plugin or npm CLI and coordinate parallel Claude Code agents through pre-built workflows instead of managing them by hand.
open-notebook
AI Assistants · Note Taking
A privacy-first, self-hosted AI research notebook with 18+ model providers, multi-speaker podcast generation, and full REST API—your open-source alternative to Google Notebook LM.
Open WebUI
AI Agents · AI Assistants
The extensible, privacy-first AI platform that runs Ollama, OpenAI, and any LLM backend behind a polished, feature-packed web interface.
OpenHands
AI Code Assistants · AI Development
The self-hosted developer control center for running AI coding agents — locally, in Docker, on VMs, or across cloud backends — with automation workflows for GitHub, Slack, and more.
OpenKB
Knowledge Management
An open-source CLI that compiles raw documents into a structured, interlinked wiki-style knowledge base using LLMs — powered by vectorless, reasoning-based retrieval (PageIndex) instead of a vector database.
opennews-mcp
AI Agents · Developer Tools · Mcp
An MCP server that plugs AI agents into 85+ real-time crypto and finance news feeds, complete with AI impact scores, trading signals, and live WebSocket updates.
OpenReplay
Analytics
Self-hosted session replay and product analytics suite that lets you see exactly what users do on your web app — without sending data to third parties.
OpenSandbox
Developer Tools · Security
Secure, fast, and extensible sandbox runtime for AI agents with multi-language SDKs and Docker/Kubernetes runtimes.