Skip to content

Evaluation Skills

Evaluation skills focus on assessing quality, verifying facts, and ensuring compliance.

SkillPurposeCommon Personas
Quality CheckVerify output quality against criteriaAda, Drucker
Security AuditAssess security posture and vulnerabilitiesRickover, Kestra
Compliance ReviewCheck adherence to standards/regulationsRickover, Ada
Fact VerificationVerify claims and assertionsWatson, Ada, Franklin

Verify output quality against predefined success criteria.

When to Use:

  • Before approving deliverables
  • Acceptance testing
  • QA reviews
  • Milestone verification

Inputs Required:

  • Output to evaluate
  • Success criteria
  • Scoring rubric (if any)

Outputs Produced:

  • Criterion-by-criterion assessment
  • Pass/fail/partial status
  • Gap identification
  • Improvement recommendations

Assess security posture and identify vulnerabilities.

When to Use:

  • Before production deployment
  • After significant changes
  • Periodic security reviews
  • Incident response

Inputs Required:

  • System or code to audit
  • Threat model (if available)
  • Compliance requirements

Outputs Produced:

  • Vulnerability list with severity
  • Risk assessment
  • Remediation recommendations
  • Compliance gaps

Check adherence to standards, regulations, or policies.

When to Use:

  • Regulatory requirements (HIPAA, GDPR, etc.)
  • Industry standards (OWASP, NIST, etc.)
  • Internal policy compliance
  • Certification preparation

Inputs Required:

  • Subject to review
  • Applicable standards/regulations
  • Previous audit findings (if any)

Outputs Produced:

  • Compliance status per requirement
  • Violation identification
  • Remediation requirements
  • Evidence documentation

Verify claims and assertions for accuracy.

When to Use:

  • Reviewing medical claims
  • Checking financial data
  • Validating research findings
  • Quality assurance on content

Inputs Required:

  • Claims to verify
  • Source documents (if available)
  • Verification standards

Outputs Produced:

  • Verification status per claim
  • Supporting evidence
  • Confidence level
  • Corrections needed