Skip to content

docs(scorecard): the verdict-of-record is asvs-scorecard.toml, not a dated prose assessment - #159

Merged
wshallwshall merged 1 commit into
mainfrom
docs-verdict-of-record
Aug 3, 2026
Merged

docs(scorecard): the verdict-of-record is asvs-scorecard.toml, not a dated prose assessment#159
wshallwshall merged 1 commit into
mainfrom
docs-verdict-of-record

Conversation

@wshallwshall

Copy link
Copy Markdown
Collaborator

One line changed, plus one added. The reason it is worth a PR is what the file already says about itself.

The defect

docs/Secure_Build_Scorecard_MEFOR.md:123 names ASVS-L3-RESCORE-2026-07-31 as "the current verdict-of-record" and instructs the reader not to grade off any earlier assessment.

ADR 0156 retired that lineage. Verdicts and evidence anchors live in docs/security/asvs-scorecard.toml, rendered to ASVS-CURRENT.md and drift-gated in CI. A prose file cannot be the record once the record is machine-checked.

Why this one is not cosmetic

The very next bullet in the same file documents this file getting the same thing wrong before. :124 reads:

ASVS-L3-ASSESSMENT-2026-07-16 — SUPERSEDED; do not cite. This entry previously called it "the current canonical verdict-of-record" and told the reader to "grade off this doc". Both were wrong by 2026-07-31…

So a dated prose pointer has gone stale here, in this file, twice — and the file is self-aware about the first instance while committing the second immediately above it. Naming the structured source is what stops a third.

This is the same class as the 15 unresolvable vault citations and the reported ~95 stale posture claims: a pointer that nothing checks, in a document nothing couples to the thing it points at.

What changed

  • Adds asvs-scorecard.toml as the record, with ADR 0156 cited.
  • Demotes the dated prose entries to provenance rather than deleting them, so a reader following an old citation still lands somewhere that explains why it is not the record.

Scope discipline

ADR 0152 deliberately untouched. It carries the same retired-lineage citation, but PR #156 already contains an ADR 0152 amendment from its author (a2a50f3d). Adding a second amendment from a different region would duplicate a decision that is not mine to record twice — it belongs with #156.

Docs only. Pre-commit leak guard and secret scan Passed; no --no-verify.

🤖 Generated with Claude Code

…dated prose assessment

This file named ASVS-L3-RESCORE-2026-07-31 as "the current verdict-of-record" and told the
reader not to grade off any earlier assessment. ADR 0156 retired that lineage: verdicts and
evidence anchors live in docs/security/asvs-scorecard.toml, rendered to ASVS-CURRENT.md and
drift-gated in CI. A prose file cannot be the record when the record is machine-checked.

The correction is small; the reason it matters is that the very next bullet in this same file
documents this file being wrong about its own pointer once already -- it says
ASVS-L3-ASSESSMENT-2026-07-16 "previously called it the current canonical verdict-of-record"
and that this was wrong by 2026-07-31. So a dated prose pointer has now gone stale here twice.
Naming the structured source is what stops a third.

Prose entries are kept, demoted to provenance rather than deleted, so a reader following an
old citation still lands somewhere that explains why it is not the record.
@wshallwshall
wshallwshall enabled auto-merge (squash) August 3, 2026 01:01
@wshallwshall
wshallwshall merged commit ab1bb88 into main Aug 3, 2026
31 checks passed
@wshallwshall
wshallwshall deleted the docs-verdict-of-record branch August 3, 2026 01:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant