What problem does ArchSteer solve?

AI agents write code faster than any architect can review, document, or govern. Docs rot instantly, the real architecture is invisible, decisions are made silently, and intended architecture drifts. ArchSteer makes architecture continuously observable and governable.

How is ArchSteer different from a linter or a diagram tool?

Linters check single-language rules; diagram tools require manual upkeep. ArchSteer maintains one code-derived model that powers living docs, auto-drafted ADRs, conformance with a net-new ratchet, an evolution feed, and AI-agent steering — all from the same source of truth.

Do I have to set up rules before I get value?

No. The read-only X-ray runs with zero config: point it at any repo and instantly see the current architecture and how it changed. Declaring intent for governance is an incremental, opt-in upgrade.

How does ArchSteer steer AI coding agents?

It injects sharp, file-scoped guardrails into CLAUDE.md / AGENTS.md / .cursor/rules that point at the governing ADR, so agents build toward the target architecture instead of copying adjacent legacy code.

Home › Architecture x-rays

Python

The architecture of HTTPX, x-rayed

HTTPX enforces its public API with a naming convention you can see straight in the x-ray: every single module is underscore-prefixed (_client.py, _models.py, _transports/). The entire supported surface is whatever __init__.py chooses to re-export — 23 private components behind one public door.

Auto-generated by archsteer xray v0.4.1 against encode/httpx@b5addb6 (httpx/) on 2026-07-05. Read-only static analysis — no code executed, no architecture rules declared. Structure, not judgment.

components

modules

internal edges

runtime deps

Module dependency graph

Top 14 modules by connectivity; an arrow means "imports from".

graph LR
    m0["__init__"]
    m1["_api"]
    m2["_auth"]
    m3["_client"]
    m4["_config"]
    m5["_content"]
    m6["_exceptions"]
    m7["_main"]
    m8["_models"]
    m9["_multipart"]
    m10["_transports"]
    m11["_types"]
    m12["_urls"]
    m13["_utils"]
    m0 --> m1
    m0 --> m2
    m0 --> m3
    m0 --> m4
    m0 --> m5
    m0 --> m6
    m0 --> m7
    m0 --> m8
    m0 --> m10
    m0 --> m11
    m0 --> m12
    m1 --> m3
    m1 --> m4
    m1 --> m8
    m1 --> m11
    m1 --> m12
    m2 --> m6
    m2 --> m8
    m2 --> m13
    m3 --> m2
    m3 --> m4
    m3 --> m6
    m3 --> m8
    m3 --> m10
    m3 --> m11
    m3 --> m12
    m3 --> m13
    m4 --> m8
    m4 --> m11
    m4 --> m12
    m5 --> m6
    m5 --> m9
    m5 --> m11
    m5 --> m13
    m6 --> m8
    m7 --> m3
    m7 --> m6
    m7 --> m8
    m8 --> m5
    m8 --> m6
    m8 --> m9
    m8 --> m11
    m8 --> m12
    m8 --> m13
    m9 --> m11
    m9 --> m13
    m10 --> m0
    m10 --> m4
    m10 --> m6
    m10 --> m8
    m10 --> m11
    m10 --> m12
    m11 --> m2
    m11 --> m4
    m11 --> m8
    m11 --> m12
    m12 --> m11
    m12 --> m13
    m13 --> m11
    m13 --> m12

Components by module

_transports · 6__init__ · 1__version__ · 1_api · 1_auth · 1_client · 1_config · 1_content · 1_decoders · 1_exceptions · 1_main · 1_models · 1_multipart · 1_status_codes · 1_types · 1_urlparse · 1

Most connected components

Top 20 of 23 by exported API surface and dependency count.

Component	Module	Key exports
`_client.py`	_client	_is_https_redirect, _port_or_default, _same_origin, UseClientDefault
`_exceptions.py`	_exceptions	HTTPError, RequestError, TransportError, TimeoutException
`_main.py`	_main	print_help, get_lexer_for_response, format_request_headers, format_response_headers
`_models.py`	_models	_is_known_encoding, _normalize_header_key, _normalize_header_value, _parse_content_type_charset
`_transports/default.py`	_transports	_load_httpcore_exceptions, map_httpcore_exceptions, ResponseStream, HTTPTransport
`_content.py`	_content	ByteStream, IteratorByteStream, AsyncIteratorByteStream, UnattachedStream
`_decoders.py`	_decoders	ContentDecoder, IdentityDecoder, DeflateDecoder, GZipDecoder
`_auth.py`	_auth	Auth, FunctionAuth, BasicAuth, NetRCAuth
`_utils.py`	_utils	primitive_value_to_str, get_environment_proxies, to_bytes, to_str
`_api.py`	_api	request, stream, get, options
`_urlparse.py`	_urlparse	ParseResult, urlparse, encode_host, normalize_port
`_multipart.py`	_multipart	_format_form_param, _guess_content_type, get_multipart_boundary_from_content_type, DataField
`_config.py`	_config	UnsetType, create_ssl_context, Timeout, Limits
`_transports/asgi.py`	_transports	is_running_trio, create_event, ASGIResponseStream, ASGITransport
`__init__.py`	__init__	—
`_transports/wsgi.py`	_transports	_skip_leading_empty_chunks, WSGIByteStream, WSGITransport
`_urls.py`	_urls	URL, QueryParams
`_types.py`	_types	SyncByteStream, AsyncByteStream
`_transports/base.py`	_transports	BaseTransport, AsyncBaseTransport
`_transports/mock.py`	_transports	MockTransport

Declared runtime dependencies

anyiocertifihttpcoreidna

What the structure teaches

One public door, 22 private rooms

Every module except __init__.py is underscore-private. That's not decoration — it gives the maintainers license to refactor anything internal without a deprecation cycle, because the only supported import path is the top-level package.

The transport layer is the extension point

_transports/ separates "what request to make" from "how bytes move": the default transport wraps httpcore, but the same interface admits mock transports for testing, WSGI/ASGI transports for calling your own app in-process, and custom ones. It's the most architecturally deliberate subpackage in the library.

Sync and async share one brain

_client.py holds both Client and AsyncClient, sharing request building, redirect logic, and cookie handling, diverging only where I/O actually happens. The x-ray shows it as the largest, most-connected component — the cost of not maintaining two parallel client codebases.

X-ray your own repo

This page is unedited archsteer xray output plus commentary. The same map of your codebase takes one command, runs locally, and never sends code anywhere: pip install archsteer && archsteer xray

Get started Use it to steer AI agents

FAQ

How is HTTPX structured internally?

As 23 underscore-private modules re-exported through a single __init__.py: _client.py (sync and async clients sharing one implementation), _models.py (Request/Response), _transports/ (the pluggable layer that actually moves bytes, wrapping httpcore by default), plus focused modules for auth, config, content, and URLs.

What is the HTTPX transport API?

A small interface (handle_request / handle_async_request) that separates the client's HTTP semantics from byte transport. Swapping it lets you test against a MockTransport, call an ASGI/WSGI app in-process without a socket, or add custom behavior — all without touching client code.

How was this architecture map generated?

By running `archsteer xray` (an open-source, MIT-licensed CLI) against the httpx/ package directory of the public GitHub repo — read-only static analysis, no code executed. Reproduce it with `pip install archsteer && archsteer xray`.