sse-starlette 3.5.0

Source: https://github.com/sysid/sse-starlette
Tag scanned: v3.5.0
Compared with: v3.4.11 (0 signals introduced since)
Files: 34 scanned, 11 skipped (binary, oversized, or unrecognised)
Signals: 0 critical · 0 warn · 1 info
Scan time: 2.5 s
LLM review: Reviewed by claude-haiku-5-5 (Anthropic): 1 of 1 warn and critical signal; 1 demoted to info.
Disclude version: 4.1.0
Scanned (UTC): 2026-10-22 02:10

Signals

Ordered by severity, then signals introduced in this release. Hidden characters are shown as ⟨U+…⟩.

info ~~critical~~ → info by review dynamic-execution tests/experimentation/test_multiple_consumers_threads.py:52:22

`subprocess.Popen` with `shell=True` and a non-literal command — runtime-determined shell invocation (ast pass, confidence 0.85)
```
    env["PYTHONPATH"] = str(ROOT_PATH)  # Set PYTHONPATH to include the project root
    server_process = subprocess.Popen(
        server_command, shell=True, cwd=ROOT_PATH, env=env
    )
```
likely benign LLM, confidence 0.85 A test helper launches a local server via subprocess.Popen with shell=True, using a project-internal command variable.
Reasoning
The snippet is a test fixture that starts a local server process with shell=True, using a command variable and a cwd and env set to the project root. Nothing in the snippet suggests attacker-controlled input or a download-and-execute chain. Shell=True with a non-literal command is a code-quality risk, but in a test harness it is most likely a developer convenience and not malicious. The only doubt is that the command's origin is not shown.

Reproduce This Scan

Run disclude locally to verify. Add --format sarif for SARIF output. Add --llm (with an API key in the environment) to include the LLM review.

cargo install disclude --version 4.1.0
git clone https://github.com/sysid/sse-starlette sse-starlette && cd sse-starlette
git checkout v3.5.0
disclude scan --severity info --diff v3.4.11 .