# CFSE Consequence Paths v1.0-candidate — independent scorer packet

This packet supports independent scoring and reproducibility work. It is a public procedure, not evidence that inter-rater reliability or construct validity has already been established.

## Inputs

For each vulnerability, collect:

1. public advisories, papers, PoCs, vendor notes, or datasets;
2. the affected product and deployment context;
3. the documented entry condition and boundary crossed;
4. any guards, scale conditions, and recovery constraints.

Do not use the existing CPATH record as an answer key when conducting an independent scoring exercise.

## Procedure

1. Write each material factual claim and attach a source.
2. Decompose distinct terminal consequences into separate paths.
3. For each path, write ordered access/boundary, capability/transition, consequence, and recovery steps.
4. Mark every step as `source`, `model-inference`, or `operational-assumption`.
5. Score TT, RE, EC, PH, DP, AT, CH, SR, SX, OR, EV, and LS. Derive EX as `min(RE, EC)`.
6. Write all twelve `metric_explanations`. For each value, give a short human label, a case-specific reason, its epistemic basis, and any source IDs it directly relies on. Do not use the compact vector as the explanation.
7. Apply the published intrinsic matrices, caps, and at most one evidence-gated systemic uplift.
8. Record a decision trail: stored base band, `none`, `cap`, or `uplift`, the case-specific reason, and the final candidate band.
9. Mark every top-band path dominant or co-dominant. Do not break ties by array order.
10. Record a plain-language dominance reason and triage implication. Preserve the compact vector as secondary machine notation for comparison and tooling.

## Submission

Use the [review template](/review-template.md), then submit through the [public review workflow](/review). Include the source set used and disclose any prior exposure to the registry result.

## Current claim boundary

The scorer packet makes independent review executable. As of v1.0-candidate, the public corpus records no completed independent path reviews, and no inter-rater or construct-validity result is claimed.