v0.6.1 of testls answers a question I've been asking: can we actually probe for known vulnerabilities, not just flag weak configurations?
--intrusive: Four checks, one flag
All four send attack traffic, so they're gated behind an explicit flag:
Heartbleed CVE-2014-0160
CCS Injection —
Ticketbleed —
ROBOT CVE-2020-14429
Findings are reported only when evidence is conclusive. Inconclusive probes are reported as not tested — never as "safe."
DH measured per suite
Diffie-Hellman group sizes are now measured per cipher suite, so the report tells you exactly which suites use weak DH parameters.
JSON schema 2 (breaking)
KeyExchangeResult::dh_prime is removed. If you parse JSON in CI, update your schemas.
Grab the static binary from GitHub Releases or
Drop it into any workflow — it supports SARIF, JUnit, Markdown job summaries, compliance policies, and baselines:
Github Actions Example:
- uses: kjpopov/testls@v0.6.1
with:
args: --fail-on medium example.com
Install via popular package managers
brew install kjpopov/tap/testls # macOS
scoop install testls # Coming Soon on Windows
Top comments (2)
Really nice improvements! Per-cipher-suite DH measurement and evidence-based vulnerability reporting are valuable additions. Also, making intrusive checks explicitly opt-in is a thoughtful security decision. Great work!
Just wired this into our staging pipeline and the --intrusive checks caught a Heartbleed-vulnerable internal service we didn't even know was still running. The per-suite DH breakdown is also way more useful than the old aggregate — we had two suites on 1024-bit DH that the previous version just lumped in. GitHub Actions integration is painless, SARIF output goes straight into code scanning. Great release.