Hacktoberfest 2026: los issues que los mantenedores marcaron para octubre, abiertos y aptos para principiantes. Explorar issues de Hacktoberfest

BUG: HTTPTarget sends conflicting Content-Length and Transfer-Encoding headers

Cerrado
  • ¿Algo no está bien?
#3,202 1 comentario 0 reacciones 0 asignados Ver en GitHub

Los mantenedores suelen responder en 2 días

@Kaap10 ya está trabajando en esto.

Desde el 6/10/2026.

  • #3008 de @Kaap10 — abierto

Evaluación

Dificultad
2/5
Tiempo estimado
1-3 horas
Aptitud para principiantes
35/100
Tipo de issue
Error
Claridad
Bien especificado
Estado de actividad
Estancado
Stack tecnológico
python
Área
api

Línea de trabajo

Start in parse_raw_http_request in pyrit/prompt_target/http_target/http_target.py, where only content-length is popped from the parsed headers. Check open PR #3008 first, since it already touches this branch and may overlap. Done means neither framing header survives parsing and the body is sent intact, which the suggested MockTransport test in tests/unit/prompt_target/target/test_request_tracing.py can pin.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

Bug: triage
Describe the bug

HTTPTarget.parse_raw_http_request strips content-length from the parsed header dict so that httpx can compute the correct value for the (possibly converter-modified) body:

https://github.com/microsoft/PyRIT/blob/main/pyrit/prompt_target/http_target/http_target.py#L290-L291

if "content-length" in headers_dict:
    del headers_dict["content-length"]

transfer-encoding is not stripped. When a raw request contains Transfer-Encoding: chunked (common when pasting from Burp), that header survives into the request, httpx/h11 chunk-frames the body, and httpx still adds its own Content-Length. Both framing headers ship on the same request.

RFC 9112 §6.1 states a sender MUST NOT send Content-Length in any message that contains a Transfer-Encoding header field. §6.3 requires recipients that receive both to treat the message as unrecoverable, and recommends intermediaries forwarding such a message strip Content-Length.

Two concrete impacts:

  1. Silent truncation / mojibake. The chunk-framed body is 20 bytes on the wire but Content-Length advertises 11 (the decoded length). Any intermediary or server that honors Content-Length reads a prefix of the chunked framing. With multibyte content this truncates mid-UTF-8 and the operator sees corrupted text rather than an error.
  2. Request smuggling shape. This is textbook CL.TE desync. PyRIT emits it with no warning, so an operator cannot tell whether an anomalous target response is a finding or a PyRIT framing artifact. For a tool whose output is security evidence, that ambiguity is the real cost.

This is pre-existing on main and unrelated to any open PR; I confirmed it reproduces both on main and on the branch in #3008.

Steps/Code to Reproduce
import asyncio, socket, threading

from pyrit.executor.attack import PromptSendingAttack
from pyrit.prompt_target import HTTPTarget
from pyrit.setup import IN_MEMORY, initialize_pyrit_async

captured = []
srv = socket.socket(); srv.bind(("127.0.0.1", 0)); srv.listen(1)
port = srv.getsockname()[1]


def serve():
    conn, _ = srv.accept()
    buf = b""
    while b"0\r\n\r\n" not in buf:
        chunk = conn.recv(65535)
        if not chunk:
            break
        buf += chunk
    captured.append(buf)
    conn.sendall(b"HTTP/1.1 200 OK\r\nContent-Length: 2\r\n\r\nok")
    conn.close()


threading.Thread(target=serve, daemon=True).start()

raw = (
    "POST / HTTP/1.1\r\n"
    f"Host: 127.0.0.1:{port}\r\n"
    "Transfer-Encoding: chunked\r\n"
    "Content-Length: 11\r\n"
    "\r\n"
    "{PROMPT}"
)


async def main():
    await initialize_pyrit_async(memory_db_type=IN_MEMORY)
    target = HTTPTarget(http_request=raw, prompt_regex_string="{PROMPT}", use_tls=False)
    await PromptSendingAttack(objective_target=target, max_attempts_on_failure=0).execute_async(
        objective="hello world"
    )
    print(captured[0].decode())


asyncio.run(main())
Expected Results

Exactly one framing header on the wire. Either transfer-encoding is stripped alongside content-length so httpx owns framing, or chunked framing is honored and no Content-Length is emitted.

Actual Results

Both framing headers are sent, and the Content-Length value does not describe the bytes actually transmitted:

POST / HTTP/1.1
host: 127.0.0.1:49678
Accept: */*
Accept-Encoding: gzip, deflate
Connection: keep-alive
User-Agent: python-httpx/0.28.1
transfer-encoding: chunked
content-length: 11

b
hello world
0

Body is chunk-framed (b\r\nhello world\r\n0\r\n\r\n, 20 bytes) while content-length: 11 describes the decoded length. No warning or error is raised.

Suggested fix

Strip both hop-by-hop framing headers in parse_raw_http_request and let httpx own framing, consistent with the existing intent for content-length:

for framing_header in ("content-length", "transfer-encoding"):
    headers_dict.pop(framing_header, None)

A regression test using httpx.MockTransport (matching the pattern in tests/unit/prompt_target/target/test_request_tracing.py) can pin that neither header survives parsing and that the body is sent intact.

If preserving operator-supplied chunked framing is considered in scope, that is a larger change and should be explicit rather than the current accidental middle state.

Screenshots

N/A

Versions
  • OS: Windows
  • Python version: 3.14
  • PyRIT version: installed from main in editable mode (1.2.0.dev0)
  • httpx: 0.28.1
Lenguaje dominante
Python
Estrellas
4.6k
Forks
944
Merge medio
2 d 18 h
PR fusionados (30 d)
369

Preparar el entorno

Abrir en Codespaces

Inicia el contenedor de desarrollo del proyecto en tu navegador, con tu propia cuenta de GitHub.

  • Sin Dockerfile ni archivo de Docker Compose
  • Tiene una plantilla de pull request
  • Sin guía de contribución

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de microsoft/PyRIT

Todos los issues de microsoft/PyRIT

Issues similares

Más issues de Python

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.