Skip to content

[REVIEW] nist-csf-assessment: separate profile scoring from CSF Tiers #93

@yunrongy424-oss

Description

@yunrongy424-oss

Skill Being Reviewed

Skill name: nist-csf-assessment
Skill path: skills/compliance/nist-csf-assessment/

False Positive Analysis

Benign scenario that can be misclassified:

An organization has strong technical implementation evidence for PR.AA-05
and DE.CM-01, but governance artifacts are weak:

- no board-approved cyber risk appetite
- no ERM integration evidence
- no repeatable oversight metrics
- no supplier participation records

Why this is a false positive:

The current skill correctly says CSF Tiers apply at the organizational level,
but Step 4 then aligns each subcategory score directly with Tier 1 through Tier
4 language. That can cause an assessor to label individual controls as "Tier 3"
or average subcategory scores into an organizational Tier. NIST frames Tiers as
organization-level cybersecurity risk governance and management characteristics,
not as per-subcategory control ratings.

Coverage Gaps

Missed variant 1: evidence confidence is not captured

GV.RM-02 = 3, Evidence = "CISO interview says appetite exists"

Why it should be caught:

Interview-only evidence should not receive the same confidence as direct board
minutes, approved policies, risk registers, or metrics. The skill needs
confidence levels and not-evaluable reason codes so assessors avoid guessed
scores when evidence is stale, third-party dependent, or out of scope.

Missed variant 2: source and mapping freshness are not tracked

Informative Refs = ISO 27001 / CIS / NIST 800-53 mapping
Source checked date = not recorded
Mapping type = not recorded

Why it should be caught:

NIST CSF 2.0 includes reference tooling, Quick Start Guides, Community Profiles,
and downloadable core/reference materials. The skill should record which source
artifact was used, when it was checked, and whether the mapping is direct,
inferred, or unavailable.

Edge Cases

  • CSF 1.1/custom IDs supplied by a client should be marked legacy/custom and
    mapped only when the CSF 2.0 source artifact supports it.
  • Supplier evidence gaps should not silently lower a control score; they should
    be marked with a specific third-party evidence reason when evidence is outside
    the assessor's reach.
  • A high technical control score should not automatically increase the overall
    organizational Tier when GOVERN evidence is weak.

Remediation Quality

  • Fix resolves the vulnerability
  • Fix doesn't introduce new security issues
  • Fix doesn't break functionality
  • Issues found: The current remediation/report format is useful, but it
    needs source-register, evidence-confidence, and not-evaluable fields to make
    the output auditable and to prevent Tier/profile conflation.

Comparison to Other Tools

Tool Catches this? Notes
NIST CSF 2.0 Reference Tool Partial Provides official CSF outcomes/reference context, but assessors still need evidence confidence and source-date discipline in reports.
NIST Quick Start Guides Partial Guides cover profiles, tiers, informative references, and community profiles, but the skill needs to operationalize those distinctions.
Manual GRC spreadsheets Partial Often track scores, but can still conflate control maturity and organizational Tier without explicit fields.

Overall Assessment

Strengths:

  • Strong CSF 2.0 structure with GOVERN, IDENTIFY, PROTECT, DETECT, RESPOND,
    and RECOVER coverage.
  • Good constraints against fabricated CSF IDs and CSF 1.1/2.0 terminology drift.
  • Useful report template and remediation roadmap.

Needs improvement:

  • Separate subcategory capability scoring from organizational Tier evaluation.
  • Add source-register fields for CSF Core/Reference Tool/Quick Start Guide or
    Community Profile artifacts.
  • Add evidence confidence and not-evaluable reason codes to avoid unsupported
    scores.

Priority recommendations:

  1. Replace the per-subcategory "Tier Alignment" scoring table with a capability
    scoring table.
  2. Add CSF-NE-* reason codes and evidence confidence levels.
  3. Add source/mapping validation and report fields for artifact date, mapping
    confidence, and target rationale.

Source Checks

Bounty Info

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions