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 FastAPI, x-rayed

FastAPI is famously a thin, sharply-focused layer: almost everything network-facing is delegated to Starlette, and almost everything data-facing to Pydantic. The x-ray makes that visible — 48 components, and the heavy ones are exactly where the framework adds its own value: routing, dependency injection, and OpenAPI schema generation.

Auto-generated by archsteer xray v0.4.1 against fastapi/fastapi@dd17e4f (fastapi/) on 2026-07-05. Read-only static analysis — no code executed, no architecture rules declared. Structure, not judgment.

components

modules

241

internal edges

runtime deps

Module dependency graph

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

graph LR
    m0["__init__"]
    m1["_compat"]
    m2["applications"]
    m3["datastructures"]
    m4["dependencies"]
    m5["encoders"]
    m6["exceptions"]
    m7["openapi"]
    m8["param_functions"]
    m9["params"]
    m10["routing"]
    m11["security"]
    m12["types"]
    m13["utils"]
    m0 --> m2
    m0 --> m3
    m0 --> m6
    m0 --> m8
    m0 --> m10
    m1 --> m7
    m1 --> m9
    m1 --> m12
    m2 --> m0
    m2 --> m3
    m2 --> m4
    m2 --> m6
    m2 --> m7
    m2 --> m9
    m2 --> m10
    m2 --> m12
    m2 --> m13
    m3 --> m0
    m3 --> m1
    m4 --> m1
    m4 --> m6
    m4 --> m9
    m4 --> m11
    m4 --> m12
    m4 --> m13
    m5 --> m1
    m5 --> m6
    m5 --> m12
    m6 --> m0
    m7 --> m1
    m7 --> m3
    m7 --> m4
    m7 --> m5
    m7 --> m6
    m7 --> m9
    m7 --> m10
    m7 --> m12
    m7 --> m13
    m8 --> m0
    m8 --> m1
    m8 --> m3
    m8 --> m7
    m8 --> m9
    m8 --> m11
    m9 --> m1
    m9 --> m3
    m9 --> m6
    m9 --> m7
    m10 --> m0
    m10 --> m1
    m10 --> m3
    m10 --> m4
    m10 --> m5
    m10 --> m6
    m10 --> m9
    m10 --> m12
    m10 --> m13
    m11 --> m0
    m11 --> m6
    m11 --> m7
    m11 --> m8
    m13 --> m0
    m13 --> m1
    m13 --> m3
    m13 --> m6
    m13 --> m10

Components by module

middleware · 7security · 7openapi · 5_compat · 3dependencies · 3__init__ · 1__main__ · 1applications · 1background · 1cli · 1concurrency · 1datastructures · 1encoders · 1exception_handlers · 1exceptions · 1logger · 1

Most connected components

Top 20 of 48 by exported API surface and dependency count.

Component	Module	Key exports
`routing.py`	routing	request_response, websocket_session, _AsyncLiftContextManager, _wrap_gen_lifespan_context
`applications.py`	applications	FastAPI
`dependencies/utils.py`	dependencies	ensure_multipart_is_installed, get_parameterless_sub_dependant, get_flat_dependant, _get_flat_fields_from_params
`openapi/models.py`	openapi	BaseModelWithConfig, Contact, License, Info
`_compat/v2.py`	_compat	evaluate_forwardref, GenerateJsonSchema, asdict, ModelField
`openapi/utils.py`	openapi	get_openapi_security_definitions, _get_openapi_operation_parameters, get_openapi_operation_request_body, generate_operation_id
`_compat/__init__.py`	_compat	—
`_compat/shared.py`	_compat	lenient_issubclass, _annotation_is_sequence, field_annotation_is_sequence, value_is_sequence
`security/http.py`	security	HTTPBasicCredentials, HTTPAuthorizationCredentials, HTTPBase, HTTPBasic
`param_functions.py`	param_functions	Path, Query, Header, Cookie
`encoders.py`	encoders	isoformat, decimal_encoder, generate_encoders_by_class_tuples, jsonable_encoder
`params.py`	params	ParamTypes, Param, Path, Query
`security/oauth2.py`	security	OAuth2PasswordRequestForm, OAuth2PasswordRequestFormStrict, OAuth2, OAuth2PasswordBearer
`exceptions.py`	exceptions	EndpointContext, HTTPException, WebSocketException, FastAPIError
`__init__.py`	__init__	—
`security/api_key.py`	security	APIKeyBase, APIKeyQuery, APIKeyHeader, APIKeyCookie
`datastructures.py`	datastructures	UploadFile, DefaultPlaceholder, Default
`utils.py`	utils	is_body_allowed_for_status_code, get_path_param_names, create_model_field, generate_operation_id_for_path
`responses.py`	responses	_UjsonModule, _OrjsonModule, UJSONResponse, ORJSONResponse
`security/__init__.py`	security	—

Declared runtime dependencies

annotated-docpydanticstarlettetyping-extensionstyping-inspection

What the structure teaches

Two files carry the framework

routing.py and applications.py are by far the most connected components — they export the APIRouter/FastAPI classes everyone touches and import from nearly every other module. This is a deliberate hub-and-spoke shape: a small public surface over many focused internal modules.

Dependency injection is a subsystem, not a feature

dependencies/utils.py is one of the largest components in the codebase. FastAPI's Depends() system — resolving nested dependencies, caching them per-request, extracting parameters — is effectively its own engine living in a two-file subpackage.

The _compat layer is how you support two Pydantic majors

The _compat module isolates every Pydantic v1/v2 difference behind one boundary, so the rest of the codebase imports compat shims instead of branching on versions inline. A textbook example of confining an external dependency's churn to a single module.

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 FastAPI structured internally?

As a thin hub-and-spoke layer over Starlette and Pydantic: routing.py and applications.py form the public hub, with focused subpackages for dependency injection (dependencies/), OpenAPI generation (openapi/), security schemes (security/), and Pydantic version compatibility (_compat/). The x-ray counts 48 components across 28 top-level modules with only 5 runtime dependencies.

What does FastAPI depend on?

Just five runtime packages: starlette (the ASGI toolkit that handles requests, responses, middleware and websockets), pydantic (validation and serialization), typing-extensions, typing-inspection, and annotated-doc. Everything else — servers, ORMs, template engines — is deliberately left to the user.

How was this architecture map generated?

By running `archsteer xray` (an open-source, MIT-licensed CLI) against the fastapi/ package directory of the public GitHub repo. It's read-only static analysis — no code is executed — and you can reproduce it on any repo with `pip install archsteer && archsteer xray`.