JANISSARY 7.3.0 Reports F1 0.921 on OWASP Benchmark, Reproducible via One Command

Version 7.3.0 of the JANISSARY security scanner has been released, with a code freeze so that it is functionally complete and protected against regressions. On the OWASP Benchmark's DAST-applicable corpus it reports an F1 score of 0.921 with precision of 1.000, and the run artifacts are committed so the result can be reproduced with a single command. The score rises from a 0.902 single-pass baseline after a re-scan of missed findings at a longer timeout, a recovery step the maintainers say was pre-registered before the run. The project also flags limitations: path traversal recall is weak at 0.049 to 0.308, so it should not be used for that class in 7.x, and an auth-gated target or single-page application can hide or distort findings. Further changes are slated for the 8.x line, while 7.x will receive only hotfixes.