Risk rating

A rule tells you a threat class applies and gives it a fixed severity. The OWASP Risk Rating Methodology asks the next question โ€” how likely, and how bad here? wyrm scores every finding Risk = Likelihood ร— Impact, seeding the factors from what the model already knows so you get a defensible number without filling in a spreadsheet.

How the score is built

Each finding carries a risk block. Likelihood and impact are each the mean of their factors on OWASP's 0โ€“9 scale, banded Low (<3) ยท Medium (<6) ยท High (โ‰ฅ6), then combined through the OWASP matrix into an overall level:

Likelihood โ†“ / Impact โ†’LowMediumHigh
LowNoteLowMedium
MediumLowMediumHigh
HighMediumHighCritical

Impact is the higher of a technical impact (driven by the CIA of the assets at stake โ€” always present) and a business impact (financial, reputation, compliance, privacy โ€” configured by the org). Business impact escalates a finding but never dilutes the technical floor.

The 16 factors, and where wyrm seeds them

Everything below is a starting point derived from the graph. It is meant to be overridden โ€” see tuning.

Threat-agent โ€” who would attack, and can they

FactorSeeded from
skill5 (advanced computer user) โ€” org knowledge; override per model.
motive5 (possible reward) โ€” org knowledge; override per model.
opportunityTrust-zone exposure: 8 when the element sits in / touches a low-trust (โ‰ค30) zone, else 4.
sizeExposure: 9 (anonymous internet) when exposed, else 5 (internal users).

Vulnerability โ€” how exploitable it is

FactorSeeded from
easeOfDiscoveryExposure: 8 when exposed, else 4.
easeOfExploitExposure: 6 when exposed, else 3.
awareness7 โ€” wyrm rules are well-known, published threat classes.
intrusionDetectionLogging control: 3 when the element is logged/audited, else 8 (OWASP scale: higher = less detected).

Technical impact โ€” what breaks if it happens

FactorSeeded from
lossOfConfidentialityMax confidentiality of the assets the element touches, scaled 0โ€“100 โ†’ 0โ€“9.
lossOfIntegrityMax asset integrity, scaled.
lossOfAvailabilityMax asset availability, scaled.
lossOfAccountabilityLogging control: 2 when logged, else 7.

Business impact โ€” what it costs the org

FactorSeeded from
financialDamage3 unless the org configures it.
reputationDamage3 unless the org configures it.
nonCompliance3 unless the org configures it.
privacyViolation7 when the data at stake is highly confidential (โ‰ฅ70), else 3.

Exposure = the lowest trust rating the element sits in (a component's zone) or touches (the min of a dataflow's two endpoints). Logging is read from an element's logging/audit/flow-log attributes, tags, or a linked logging mitigation.

Tuning โ€” three layers, last wins

  1. Model seed โ€” the factors above, straight from the graph.
  2. Org config โ€” .threatmodel/risk.yaml sets the factors the graph can't infer (attacker profile, business impact), applied to every finding.
  3. Per-element attributes โ€” risk.<factor> on a component or dataflow overrides just that element (0โ€“9, clamped).

.threatmodel/risk.yaml

Org-wide defaults. Every key is optional; anything unset keeps its model-derived seed.

# Threat-agent profile for your context
skill: 6            # e.g. targeted by capable, funded actors
motive: 7
size: 9             # your surface is internet-facing by default

# Business impact โ€” what a breach costs *this* org
financialDamage: 7
reputationDamage: 8
nonCompliance: 6    # regulated (PCI, GDPR, โ€ฆ)
privacyViolation: 8

Per-element overrides

Any of the 16 factors, as a risk.<factor> attribute (camelCase names), on the element it applies to:

components:
  - id: payments-api
    name: Payments API
    type: web-service
    parent: { trustZone: edge }
    assets: { processed: [card-data] }
    attributes:
      risk.easeOfExploit: "2"   # hardened, pentested endpoint
      risk.financialDamage: "9" # money moves through here

Reading the score

wyrm analyze prints the rating under each finding; the GUI findings panel shows an OWASP badge (hover for the likelihood ร— impact breakdown); and SARIF export carries the full block as an owaspRisk property so it survives into GitHub code scanning and downstream tooling.

HIGH  Information Disclosure ยท Payments API ยท WYRM-T008
      โ†ณ Require authn/authz in front of this endpoint.
    risk: Critical (likelihood 7.0 High ร— impact 7.3 High)

The rule's severity says this class of threat is serious; the OWASP level says and in this position, with this data, it's a Critical you should fix first. Sort by the risk level to triage.