Edward A. Neyhart

Step 3 of 7 · How he works

The loop, run as an incident

Cucumbs describes its work as a nine-step loop. Incident response runs on the same loop, and each step below has evidence from Edward's record.

Root: axiom VI Flow is intelligence in motion. The pairing is this site's own reading.

Observe, understand, decompose, delegate, compare, reconcile, synthesize, act, recurse

Each step has three parts: what the step means, how it plays out in an incident, and where it shows up in his career. The incident lines describe general practice. They aren't taken from a specific case. The hub has its own loop walkthrough if you want the general version.

  1. Observe

    What it means
    Watch what the environment is actually doing before deciding what it means.
    In an incident
    Alerts, logs and hunting queries show something off on a handful of hosts.
    On the record
    FACT Built KQL analytics rules and hunting queries in Microsoft Sentinel at SecurIT360. At Equifax, detected Crimea-linked activity before it was public news.
  2. Understand

    What it means
    Work out how the technology, people, processes and policies fit together.
    In an incident
    Before touching anything, map what the affected systems connect to and who depends on them.
    On the record
    FACT Akamai engagements started with a posture assessment.
  3. Decompose

    What it means
    Separate the underlying problem from the symptoms it produces.
    In an incident
    Split the incident into what is happening, how it got in, and what has to stop first.
    On the record
    FACT Was the go-to person at Akamai for troubleshooting and root-cause analysis.
  4. Delegate

    What it means
    Give each piece to the people who will run it, with clear written instructions.
    In an incident
    Containment, evidence handling and communication go to the people who own them, each with a runbook.
    On the record
    FACT Wrote the runbooks for each Sentinel build and handed it to SOC operations. Has trained and mentored people since 1997.
  5. Compare

    What it means
    Test the options side by side before committing to one.
    In an incident
    Weigh containment options against what each one would break.
    On the record
    FACT Ran vendor selection and bake-offs at Vigilar, Sprint, EarthLink and Equifax.
  6. Reconcile

    What it means
    Settle the conflict when a product, a policy and the real environment disagree.
    In an incident
    When a tool and the environment disagree, find out why rather than overriding one.
    On the record
    FACT Worked with Check Point and Carbon Black engineering teams to fix product bugs and integration problems in production. Owned the PCI-DSS Level 1 program at EarthLink.
  7. Synthesize

    What it means
    Pull the pieces into one picture that people can act on.
    In an incident
    Write the after-action report and brief leadership on the business risk as well as the technical detail.
    On the record
    FACT Wrote after-action reports at SecurIT360. Briefed Equifax leadership on business risk and wrote the Security Engineering and Assessment Standards.
  8. Act

    What it means
    Make the change, fast where it has to be and careful where it can be.
    In an incident
    Contain, then put lasting controls in place.
    On the record
    FACT Average containment of 60 to 90 minutes across 300+ incidents a year at SecurIT360. Put geo-fencing in place at Equifax. Ran a zero-downtime migration at Sprint.
  9. Recurse

    What it means
    Feed what was learned back into the start of the loop.
    In an incident
    The root cause becomes a tuned detection and an updated playbook, so the next attempt is caught at step one.
    On the record
    FACT Fed root cause back into detections and playbooks at SecurIT360. That cut false positives by about 35%. Wrote lifecycle processes at the State of Georgia that teams used for later integrations.

The ownership arc

Across his career, Edward has owned projects the same way from start to finish. He doesn't hand off a design and walk away. He stays with it until the team running it no longer needs him, and then stays on as their escalation point.

  1. Vendor selectionPick the candidates worth testing.
  2. Bake-offTest them against each other in real conditions.
  3. DesignPlan how the chosen tool fits the environment.
  4. ImplementBuild it and put it into service.
  5. DocumentWrite the runbooks and playbooks.
  6. Hand offPass it to the operations team that will run it.
  7. Train and mentorTeach the people who run it day to day.
  8. Senior escalationStay the person they call when it breaks.

How he thinks about systems

His technical background is the evidence. The habit underneath it is reading an environment as one connected system.

  • Root cause over symptoms

    He asks why a problem keeps coming back, instead of treating the same consequence again.

  • People are part of the system

    He looks at how people, automation, policy and training interact, including when a system that looks efficient on paper asks something unrealistic of the people running it.

  • Patterns across domains

    A structural problem in one industry often shows up in another. He carries the lesson across.

  • Practical changes

    Analysis has to end in a change someone can make and measure. He is an engineer, and he wants the analysis to end in a fix.

How this page was built

Sections

  • Observe, understand, decompose, delegate, compare, reconcile, synthesize, act, recurse
  • The ownership arc
  • How he thinks about systems

Sources

  • Edward's career record
  • The Cucumbs loop

Interactive part

Loop stepper.

It runs in your browser, sends nothing and stores nothing. With JavaScript off, the page still shows everything.

Checks before review

Banned-term scan, inline-code scan, outside-link check, HTML parse, a local serve, and screenshots on desktop and mobile with JavaScript on and off.

The full build record