About Cognatory

Independent security consulting

Cognatory brings a researcher’s perspective to security questions: identify the assumptions, examine the evidence and verify what changes. Engagement scope should follow the system and the decision at hand. A useful outcome is a clear explanation, an actionable improvement and evidence that supports the conclusion.

Research, applied

OpenSSH’s 10.5 release notes credit Brian Mingus of Cognatory with the report and fix for the client use-after-free associated with CVE-2026-73282. This is a specific, independently documented contribution. The research note explains what the public sources establish and where the claim ends.

Human judgment and modern tools

Cognatory uses Antigravity CLI, Codex CLI and Desktop, GitHub Copilot with Claude, and Cursor. The purpose is to accelerate investigation and engineering while keeping scope, evidence and consequential decisions explicit. Tool output remains something to inspect and test.

Brian’s research background spans learning, cognition and computational systems. That perspective informs the questions we ask about software and AI; it does not replace an assessment of the system in front of us.

About the resource library

The security guides are published under Cognatory Editorial. They provide practical decision aids, clearly identified illustrative examples and links to primary references. The initial library is introductory: listing a subject does not imply exhaustive control coverage, specialist accreditation or a claim that a particular client engagement occurred.

Published dates identify the edition of each guide. Check current product documentation and requirements before applying it.

Start with your question

Describe the system, the question you need answered and any operational constraints. Avoid including passwords, private keys or sensitive customer records in an initial message. We can establish the appropriate scope and handling arrangements before examining that material.