AgentDiscover Scanner v2.6.0 — Find the AI You Don't Know About #6
mwaseem-defendai
announced in
Announcements
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
AgentDiscover Scanner v2.6.0
What's new:
--summaryflag: executive-readable output for CISO reporting--src-repoflag: scan additional git repos as supplementary source layerGhost agent detection — runtime AI activity with no source code —
remains the core capability. v2.6.0 makes the first scan useful on
every platform, not just Linux+K8s.
Full changelog: CHANGELOG.md
Why this release matters
The most common feedback from early users was silent failures —
running
scan-all ~/projectson a macOS dev machine and seeing atable of zeros. No findings, no explanation, no signal that anything
was working.
The root cause: GHOST detection requires correlating runtime network
activity against source code. On a macOS machine without K8s, the
runtime layer had almost nothing to work with. The scanner wasn't
broken — the environment just wasn't producing the signals it needed.
But users couldn't tell the difference.
v2.6.0 fixes this by making the first scan useful on every platform.
Cursor, Claude Desktop, Ollama — detected instantly via process
scanning before the network observation window starts. They surface
as SHADOW AI with governance recommendations rather than disappearing
into a clean result that means nothing.
What we want to hear from you
catching? Open an issue with the process name.
we can get.
unusual K8s distro)? Tell us what breaks.
This discussion was created from the release AgentDiscover Scanner v2.6.0 — Find the AI You Don't Know About.
All reactions