Skip to content

docs(adr): record check precision defaults in ADR 0080, amending ADR 0003 - #8707

Merged
kyle-sexton merged 2 commits into
mainfrom
work-resolve-issue/issue-8537-39474783153-1
Oct 21, 2026
Merged

kyle-sexton merged 2 commits into
mainfrom
work-resolve-issue/issue-8537-39474783153-1

Conversation

@melodic-automation-lanes

Copy link
Copy Markdown
Contributor

Closes #8537

Summary

Opened by the work-resolve-issue lane. The run's own summary:

Adds ADR 0080, which amends ADR 0003 with the precision defaults for checks: 0 effective false positives for a blocking check, under 10% for an advisory one (named a corpus-precision proxy), plus the exception rules. ADR 0003 is not edited.

Not verified

  • scripts/check-adr-numbers.sh --check was not run here: no wrapper for it is available in this run, and Bash listing of docs/adr was denied. I took 0079 as the highest number from the issue and did not confirm it independently. check-affected lists adr-numbers as CI only; left to the PR's lint-repo j
  • Markdown lint was not run locally; left to the PR's lint-repo job.
  • Code review of the diff was not run in a subagent. The change is one new documentation file.

Fix

  • docs/adr/0080-set-check-precision-defaults-at-zero-false-positives-blocking-and-under-10-percent-advisory.md: new ADR with an Amends line for ADR 0003, Context, Decision and Consequences

Verification

The runner ran each check from the base copy against 715fe3d:

  • scripts/check-changelog-fragments.sh --check: passed

The run reports:

Left out: the items under Not verified.

Related

#8537

…0003

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Oct 20, 2026 •

Copy link
Copy Markdown
Contributor

claude-security-review needs no new review of 982318466f4b46ffd5760456c996575a5fd1023f: every file in scope matches docs-only-paths.

@github-actions

github-actions Bot commented Oct 20, 2026 •

Copy link
Copy Markdown
Contributor

claude-review reviewed 982318466f4b46ffd5760456c996575a5fd1023f: 0 finding(s), 0 as inline comments.

Review: incremental (715fe3d → 9823184)

Scope: the one changed file, ADR 0080. I found nothing to report.

What changed and how I checked it

  • Effective false positive: the definition now matches the Tricorder wording. A dismissal, or a clearance by exception instead of a fix, counts. The new break-glass sentence agrees with it, and the two don't conflict.
  • Ladder claim: the Consequences text now says the guardrail criterion "will read" these numbers and that the ladder has none yet. A grep of docs/conventions/enforcement-ladder/README.md for guardrail, precision and 0080 found no matches, so the hedge is accurate. It also fixes the earlier overstatement that the ladder already reads the numbers.
  • Settings sentence: "No plugin setting or script reads feedback.disable_not_useful_ratio or exceptions.max_days yet" is a softer claim than the old "exist only in this decision record". This diff does not add either name outside the ADR.
  • Formatting: the long lines in the Context paragraph and the ## Sources entries are not a lint problem. .markdownlint-cli2.jsonc sets MD013: false. The new [Tricorder](#sources) anchor matches the new ## Sources heading.
  • Closing issue docs(adr): record check precision defaults in an ADR that amends ADR 0003 #8537: the ADR still contains 0 effective false positives, 10%, 90 days and #3756, and still has its Amends: line for ADR 0003. The edits don't affect the other acceptance criteria. Two of them can't be settled from the diff:
    • check-adr-numbers.sh (lint-repo): I did not run it.
    • "Diff touches only docs/adr/": the incremental diff itself touches only this ADR.

Unverified

  • I did not fetch the Tricorder PDF. The three quotes in ## Sources ("any report from the tool where a user chooses not to take action to resolve the report", "essentially zero", "< 10%") and the claim that the paper measures its 10% as a not-useful click rate are unchecked against the source. A maintainer who wants to confirm them should read the paper.
  • The PR body's own verification claims are the author's and unconfirmed.

@melodic-automation-lanes

melodic-automation-lanes Bot commented Oct 20, 2026 •

Copy link
Copy Markdown
Contributor Author

Unattended merge: risk MED at head 982318466f4b46ffd5760456c996575a5fd1023f. Only LOW merges without a human. This pull request waits for one.

MED evidence-unbound:claude-security-review: not bound: skip-not-no-change

This refusal has stood unchanged at this head, so a person is needed: the pr-merge lane added the human-gated label.

Define an effective false positive in the Tricorder paper's sense: a
finding the author dismisses or clears with an exception instead of
fixing counts. Each break-glass exception against a blocking check
counts as one. Add a Sources section quoting the paper's definition and
its essentially-zero and under-10% bars, put the ladder Consequence in
future tense, and drop the file-inventory claim this ADR falsified.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@melodic-automation-lanes melodic-automation-lanes Bot added the needs-human Human-in-the-loop required; autonomous sessions must not resolve items carrying this. label Oct 20, 2026
@kyle-sexton
kyle-sexton added this pull request to the merge queue Oct 21, 2026
Merged via the queue into main with commit 8059651 Oct 21, 2026
39 checks passed
@kyle-sexton
kyle-sexton deleted the work-resolve-issue/issue-8537-39474783153-1 branch October 21, 2026 01:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

needs-human Human-in-the-loop required; autonomous sessions must not resolve items carrying this.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

docs(adr): record check precision defaults in an ADR that amends ADR 0003

1 participant