Guard Core flaw disables two newline-based detectors before version 3.13.0

  • Guard Core
  • WAF
  • preprocessing
  • detection bypass
  • pipguard-core

Whitespace preprocessing previously removed critical newline characters

Update to version 3.13.0 restores HTTP splitting and command injection detection

The flaw affects pipguard-core 1.0.0 through versions before 3.13.0

Guard Core’s default “enhanced” detection mode has a flaw that disables its HTTP-splitting and command-injection detectors in pipguard-core versions >= 1.0.0 and < 3.13.0. The GitHub advisory says the issue stems from preprocessing that removes newline characters before the software checks input against detection patterns.

The http_split detector looks for a carriage return or line feed, while the cmd_injection detector requires a line feed. Because preprocessing collapses all whitespace into spaces, neither detector can match its required newline in the scanned value.

The flaw can leave HTTP response-splitting-shaped or newline-anchored command-injection input unflagged by this detection layer. The advisory says this is not directly exploitable against applications using Starlette and uvicorn, which reject carriage returns and line feeds in response headers; risk may remain in applications using other stacks or re-emitting stored input where a parser accepts those characters.

The advisory lists version 3.13.0 as patched. It recommends preserving carriage returns and line feeds during whitespace collapse or checking the original content with the newline-anchored patterns.