Skip to main content

attackmap-analyzer-python

Comprehensive Python ecosystem analyzer for AttackMap.

This plugin is additive over AttackMap's built-in python-web analyzer. The built-in handles FastAPI / Flask routes via the core scanner; this plugin adds everything else.

  • Web frameworks —
    • Django routing (path("users/", ...), path("", ...) → /, re_path(r"^api/...$", ...), legacy url(...) / patterns(...)) gated by urlpatterns = in the file, with the include() graph resolved repo-wide: include("app.urls"), include(module), include((patterns, "app")), include([...]) and local pattern lists join their prefix onto every child route (path("api/", include("shop.api_urls")) + orders/<int:pk>/ → /api/orders/<int:pk>/)
    • Django REST Framework routers (router.register(r'users', UserViewSet), any *Router()), mounted under their include(router.urls) / urlpatterns = router.urls prefix, and @api_view(['GET', 'POST']) decorators
    • Starlette Route("/x", endpoint=h, methods=["GET", "POST"]) and WebSocketRoute("/ws/...")
    • AIOHTTP app.router.add_get/post/..., add_route("POST", ...), add_view, web.get/post/view(...) helpers and web.RouteTableDef() decorators (@routes.get("/x"), @routes.view, @routes.route("PATCH", ...)) (gated by aiohttp import)
    • Sanic @app.get("/x"), @app.post("/x"), @app.route("/x", methods=[...]), @app.websocket("/ws") (gated by sanic import)
    • Litestar (and legacy Starlite) handlers at any indentation (@get(), @post(path="/x"), @route(http_method=[...]), @websocket), Controller classes (path = "/users" joined onto each handler) and Router(path=..., route_handlers=[...]) prefixes, including nested routers across files (decorators must be imported from litestar/starlite). guards=[...] on the handler/controller/router sets the route's auth to required; exclude_from_auth=True sets it to anonymous
    • Tornado Application([(r"/x", Handler), url(r"/y", H)]) URL specs paired with the handler class's get/post/... methods (a WebSocketHandler is WS); @tornado.web.authenticated on a method sets its auth to required
    • Flask app.add_url_rule("/x", view_func=...) (additive — built-in covers @app.route)
  • Databases —
    • SQLAlchemy create_engine("postgresql://...") with dialect-aware kind inference (postgresql / mysql / mariadb / sqlite / oracle / sqlserver)
    • asyncpg / psycopg2-3 / pymysql / motor.motor_asyncio / pymongo / redis.asyncio / aioredis / sqlmodel / tortoise
    • boto3 resources/clients for dynamodb and s3
    • Django DATABASES dict in settings.py — ENGINE value parsed for the kind
  • Auth packages —
    • passlib CryptContext, passlib.hash.argon2/bcrypt/scrypt
    • bcrypt standalone (bcrypt.hashpw, bcrypt.checkpw)
    • argon2-cffi (from argon2 import PasswordHasher)
    • PyJWT / python-jose (jwt.encode, jwt.decode, jose.jwt)
    • authlib (OAuth(...), OAuth2Session(...))
    • fastapi-users (BearerTransport, JWTStrategy)
    • flask-jwt-extended (@jwt_required, create_access_token)
    • django.contrib.auth (authenticate, make_password)
    • casbin (Enforcer)
  • HTTP clients (external calls) — httpx sync + async, aiohttp.ClientSession, urllib.request.urlopen
  • Secrets —
    • os.environ.get("X") and os.environ["X"] with secret-shaped names (built-in only covers os.getenv)
    • pydantic-settings BaseSettings field-name extraction (jwt_secret: str = ...)
    • dotenv_values(...) accessors with secret-shaped keys
  • Service hints — [project] name from pyproject.toml, [tool.poetry] name, name="..." from setup.py, [metadata] name from setup.cfg, presence of manage.py → framework:django, Django INSTALLED_APPS third-party / local apps

All emissions populate AttackMap's Signal v2 fields (line numbers, evidence snippets, confidence) so downstream insights can cite path/to/module.py:NN.

Install

uv tool install attackmap --with attackmap-analyzer-python    # from PyPI
# or: pip install attackmap-analyzer-python    (installs attackmap too)
# or from source:
pip install git+https://github.com/mlaify/attackmap-analyzer-python.git

The analyzer is auto-discovered by AttackMap via the attackmap.analyzers entry-point group.

Usage with AttackMap

# Auto-discovered when installed:
attackmap analyze /path/to/python/repo

# Or invoke explicitly:
attackmap analyze /path/to/python/repo --module python

Detection

detect() returns true when any of the following are present, ignoring .venv/, venv/, env/, .tox/, __pycache__/, .pytest_cache/, dist/, build/, node_modules/, .git/, .mypy_cache/, .ruff_cache/, and site-packages/:

  • A pyproject.toml, setup.py, setup.cfg, manage.py, or Pipfile at the root
  • Any .py file in the tree

Coverage notes

  • Framework-gated extraction: route extractors for AIOHTTP, Sanic, Litestar, and Starlette only fire when the file imports the corresponding framework. This prevents mis-attributing generic @get(...) / web.get(...) / app.get(...) calls to a framework when they're really custom decorators or method calls.
  • Django routing requires urlpatterns = to be in the file. A bare path("/x", ...) in any other context (e.g., pathlib.Path("/x") is well-behaved, but a project that also has from django.urls import path; path("/x", view) outside a urls.py won't fire — by design).
  • Django settings extraction: gated by presence of one of INSTALLED_APPS, DATABASES, MIDDLEWARE, or ROOT_URLCONF. Built-in Django apps (django.*) are filtered out of the INSTALLED_APPS service hints to keep signal high.
  • DRF @api_view(['GET', 'POST']): emits framework hints (drf_api_view:GET, drf_api_view:POST) rather than routes — the actual path is in urls.py, which is extracted separately.
  • Django include() resolution: a module reference resolves only to a URLconf inside the repo (tried from each manage.py directory, the repo root, then the including file's ancestors). A resolved include() line is a prefix, not an endpoint, so it isn't a route. An include that can't be resolved (a third-party app, include(get_model_urls(...))) is still emitted as its prefix (/accounts/). A pattern list or router nothing mounts keeps the unprefixed emission. A non-literal route (path(settings.BASE_PATH, include(...))) joins as an empty prefix. Regex anchors are stripped when joining, so re_path(r"^users/$") is /users/. A pattern list is the union of everything assigned to it (reassignments, if/else and try/except branches, urlpatterns = wrap(urlpatterns)), so no branch's routes are dropped. The resolver is bounded (5k entries per list, 20k per file, 5k registrations per router, 12 include levels, 64 mounts per list, 200k steps including router registrations, 50k routes) and cycles stop.
  • Parsing: URLconfs, Litestar, Tornado and Sanic files are read with ast (never executed). A file ast can't parse (Python 2, over 1 MB) falls back to the per-file regex extractors.
  • Not yet: DRF nested routers' parent prefix, zulip-style custom rest_path(...) wrappers, Sanic Blueprint(url_prefix=...) joins (core doesn't join them either, so joining here would duplicate routes), and Django route auth decided at the view (@login_required on the view function, LoginRequiredMixin, DRF permission_classes) — see the issue tracker. Auth wrappers applied in the URLconf itself (login_required(views.x), permission_required("p")(View.as_view()), staff_member_required, user_passes_test) do set the route's auth to required, and Django 5.1's login_not_required(...) sets it to anonymous.
  • pydantic-settings field extraction matches any class field whose name contains secret, token, key, password, pass, or pwd. False positives are possible (e.g., a non-secret signing_key config). Confidence is the standard 0.85.

Coexistence with the built-in python-web analyzer

AttackMap ships with a python-web built-in that handles FastAPI / Flask via the core scanner. When this plugin is installed, both analyzers run on every Python repo. AttackMap's overlay deduplication merges their output so users see a single unified set of findings — no double-counting.

This plugin runs at priority=10. AttackMap runs analyzers in (priority, name) order across built-ins and plugins and merges first-seen-wins, so this plugin runs before the built-in (priority=20) and its richer output wins where they overlap.

License

MIT

Metadata

Release files for attackmap-analyzer-python 0.2.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for attackmap-analyzer-python 0.2.1
File Size Uploaded Provenance Actions
attackmap_analyzer_python-0.2.1.tar.gz 47.9 kB With provenance Details for attackmap_analyzer_python-0.2.1.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for attackmap-analyzer-python 0.2.1
File Interpreter ABI Platform Actions
attackmap_analyzer_python-0.2.1-py3-none-any.whl Python 3 none any Details for attackmap_analyzer_python-0.2.1-py3-none-any.whl

Total release size: 82.6 kB

Release files / attackmap_analyzer_python-0.2.1.tar.gz

Download URL attackmap_analyzer_python-0.2.1.tar.gz
Size 47.9 kB
Tags Source
SHA-256 checksum
How to use checksums
7c646caeb9bb7444f4917f03084b6d985c75b9a67b27e61fd8f3c495ba4e5342
BLAKE2b-256 checksum
How to use checksums
fadc38df1b18823f30945b6e7c35437b98908a787c343a9374e383d56d9b38e3
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 14, 2026.

Transparency log

Release files / attackmap_analyzer_python-0.2.1-py3-none-any.whl

Download URL attackmap_analyzer_python-0.2.1-py3-none-any.whl
Size 34.8 kB
Tags Python 3
SHA-256 checksum
How to use checksums
49567703ba6ca9d2e6427d1ef1b1f59dbcf72463891de7103d99aa06dc164e30
BLAKE2b-256 checksum
How to use checksums
257af70e2a0d90a12b83e74f5a5c010626e4398bf2c97e8f70802bdceab0989b
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 14, 2026.

Transparency log

Security for attackmap-analyzer-python 0.2.1

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for attackmap_analyzer_python-0.2.1.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 14, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.2.1 This release

2 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page