Five findings, each rated low. Together, one path into production.
Coldbraid connects the signals security tools report in isolation, verifies the path they form with read-only checks, and hands you the change that breaks it.
Example path of five findings that converge into one attack path when connected.
old-staging.halden.example still resolves to a live host.
Informational
Directory listing enabled at /exports/.
Low
deploy-2019.env in that listing contains an access key.
Low, archived 2019
Key AKIA…7Q last rotated in 2019. Still active.
Low, rotation policy
halden-prod-backups grants read to that key’s user.
Low, internal principal
Coldbraid connects the signals security tools report in isolation, verifies the path they form with read-only checks, and hands you the change that breaks it.
A finding is a fact about one asset. A path is what several facts mean together. Coldbraid links each finding to the assets it touches and to the findings it makes reachable, then follows the chain to where it ends: a database, a bucket, a mailbox, a build system.
The five findings above come from five different tools. None of them is wrong. None of them, alone, says anything about production backups.
A path on paper is a hypothesis. Coldbraid confirms each link with read-only checks: a hostname that resolves, a listing that returns, an identity call that answers. Nothing is written, deleted or degraded along the way.
A path that fails verification is reported as exactly that, so a team never spends a week on a route an attacker could not take.
Every verified path ends in the change that breaks it, placed at the link where the change costs least. In the path above: rotate one key, and the other four findings go back to being low.
Fixes arrive as the exact change to make, in the language of the system that needs it, not as a page in a report.
Coldbraid is in development. Leave a work address and we will write once, when there is something to show.
