Cogent Verification And Closure Validation
Cogent Platform & Cogent Community · Cogent Security
Agentic AI Vulnerability Management — Cogent Security
Cogent Security evals — Verification & Closure Validation (relift v3 InfraRed)
About Cogent Security
Cogent Security builds agentic AI for vulnerability management. The Cogent Platform runs Triage, Risk Assessment, Remediation, and Verification agents on a real-time data foundation — investigating findings, correlating assets to owning teams, prioritizing by real exploitability over raw CVSS, driving remediation through engineering workflows, and validating that fixes actually happened. The free Cogent Community surface pairs VulnCheck-powered CVE intelligence with a customizable Discover Feed and an AI Research Assistant that produces cited, plain-language deep-dives.
Employees
~30
Industry
AI Security / Vulnerability Management
Headquarters
San Francisco, CA
Website
www.cogent.comSample tests· showing 3 of 9
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | Engineer marks a remediation ticket as 'closed — patched'. The next scanner run still finds the vulnerable version on 3 of the 12 hosts. | Per Cogent's documented principle ('validates that remediation actually happened, not just that tickets closed'), the Verification step must reject the closure, reopen the finding scoped to the 3 affected hosts, and surface the gap to the engineer with the scanner evidence. | Pass / FailAi Platformcritical |
| 02 | Remediation is a config change (e.g., disable insecure cipher in TLS config). Engineer asserts done. | Verification must reach the live config — either via active probe of the endpoint (TLS handshake) or via the host config telemetry — and confirm the cipher is disabled. Closure must not rely on the engineer's assertion. | Pass / FailAi Platformhigh |
| 03 | Patched binary installed; the running process is still the pre-patch version because the service was not restarted. | Per the open question on distinguishing 'patch applied' from 'patch applied but ineffective', Verification must check process-runtime state (e.g., loaded library version) — not just the installed package version — and refuse closure until the service runs on the patched binary. | Pass / FailAi Platformcritical |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Cogent
- Ai Platform
- Verification And Closure Validation
Recommended for
Works with
Related evals
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.