Guide · Detection engineering

Turning coverage gaps into detection engineering work

Last reviewed:

A coverage gap is not a ticket until the team knows what evidence is missing, what fix is possible and who can ship it.

Name the missing evidence

Start by naming what is absent. Is there no log source, no parser, no field mapping, no rule, no triage path or no owner? Each answer leads to different work.

This matters because 'missing T1078.004 coverage' could mean Entra audit logs are not ingested, AWS IAM events are not normalised or the team has a rule but no alert route. One label cannot fix all three.

Choose the fix type

Turn the gap into one of a few fix types. Telemetry gaps need logging or ingestion work. Logic gaps need Sigma or SIEM rules. Context gaps need asset, identity or exposure data. Workflow gaps need ownership and triage steps.

That split keeps detection engineering from absorbing every cloud security problem. Some gaps belong with platform teams before a detector can be useful.

  • Telemetry fix: enable or ingest the required events.
  • Rule fix: write, convert and test detection logic.
  • Context fix: add asset, identity or exposure data for ranking.
  • Workflow fix: define triage, ownership and escalation.

Rank work by path

Rank the queue by attack path, exposure and delivery cost. A gap for internet-facing initial access deserves attention before a niche post-exploitation technique in an isolated lab account.

CloudSigma fits the rule-fix lane once the telemetry is real and the behaviour is specific. DCV fits the review lane by keeping the missing evidence and the fix type visible.

Write actionable tickets

A useful ticket should tell the engineer what to build and how the reviewer will know it worked. If the ticket says only 'improve coverage', it is not ready.

Add the ATT&CK technique, platform, missing evidence, proposed fix, validation check and owner. The ticket can be small; vague work is the expensive part.

Gap ticket shape text
technique: T1562.008
platform: AWS
missing_evidence: CloudTrail coverage for GuardDuty disablement path
fix_type: SIEM detection rule
validation: replay known DisableDetector event in test account
owner: detection-engineering
Sources
  • MITRE ATT&CK Enterprise matrix, https://attack.mitre.org/matrices/enterprise/
  • CISA Known Exploited Vulnerabilities catalog, https://www.cisa.gov/known-exploited-vulnerabilities-catalog
  • Google Cloud Security Command Center findings, https://cloud.google.com/security-command-center/docs/concepts-security-sources
  • AWS GuardDuty finding types, https://docs.aws.amazon.com/guardduty/latest/ug/guardduty_finding-types-active.html
Last verified: 2026-05-20