Impact
detectThreat in guardcore/pipeline.go scans every value of a request (path, query parameters, headers, body) for penetration patterns. When the detector reported a threat for a value but none of that value's threats carried a reportable category, the scan returned a clean verdict for the whole request instead of moving on to the next value. Two cases trigger it:
- a semantic-only threat (semantic threats carry no category), and
- a threat whose categories are all disabled or skipped for that component (for example
ssrf on the excluded proxy headers).
Every value scanned after that one, including later headers and the request body, was never checked. A request that puts such a value early can carry an attack payload in a later value without it being detected or blocked.
Patches
Fixed in v4.4.0. A value whose threats are all filtered out is no longer a threat and the scan continues with the next value; a semantic-only threat is reported (with its attack type as the category), matching the Python reference engine.
Workarounds
None that keep detection enabled. Upgrade to v4.4.0.
Impact
detectThreatinguardcore/pipeline.goscans every value of a request (path, query parameters, headers, body) for penetration patterns. When the detector reported a threat for a value but none of that value's threats carried a reportable category, the scan returned a clean verdict for the whole request instead of moving on to the next value. Two cases trigger it:ssrfon the excluded proxy headers).Every value scanned after that one, including later headers and the request body, was never checked. A request that puts such a value early can carry an attack payload in a later value without it being detected or blocked.
Patches
Fixed in v4.4.0. A value whose threats are all filtered out is no longer a threat and the scan continues with the next value; a semantic-only threat is reported (with its attack type as the category), matching the Python reference engine.
Workarounds
None that keep detection enabled. Upgrade to v4.4.0.