Skip to content

Limit RFC 0020's later-sibling rule to admitted instances - #240

Merged
kikashy merged 3 commits into
mainfrom
rfc0020-213-209
Oct 22, 2026
Merged

kikashy merged 3 commits into
mainfrom
rfc0020-213-209

Conversation

@kikashy

@kikashy kikashy commented Oct 22, 2026

Copy link
Copy Markdown
Contributor

Closes #213.

What it adds

  • Qualifies RFC 0020 §2 so the later-sibling limit-exceeded rule applies only to instances admitted for acquisition, while §§3 and 7 admission failures retain their own codes.
  • Adds a closed, hand-written workflow vector with the four ordered instances from the decision and a negative twin whose incorrect projection-failure code produces inventory-mismatch.
  • Updates the companion, RFC fixture plan, workflow fixture schema and documentation, tests, and changelog without adding a workflow instance member or state value.

Checks

== check summary (dev.json) ==
handover: boundary 2 paths, shared 3 paths; see AGENTS.md "Before you hand over"
PASS     local-paths              0s
PASS     agent-docs               0s
    sync-agent-docs: CLAUDE.md matches AGENTS.md
PASS     diff-check               0s
PASS     repository-tests         105s
    Ran 245 tests in 104.851s
    OK
PASS     policy-prototype         1s
    Ran 38 tests in 1.529s
    OK
PASS     policy-contract          3s
    Ran 13 tests in 3.091s
    OK
PASS     policy-official-draft    3s
    Ran 16 tests in 3.671s
    OK
== 7 passed, 0 failed, 0 skipped, 0 not run ==

Not in this PR

A Codex session wrote this change.

kikashy and others added 2 commits October 22, 2026 15:08
Qualify the limit-exceeded later-sibling rule so projection and influence
admission failures retain their own codes. Add hand-written positive and
negative workflow vectors that exercise both rules and the verifier finding.

Co-Authored-By: Codex <noreply@openai.com>
Signed-off-by: kikashy <kikashy@hotmail.com>
@kikashy

kikashy commented Oct 22, 2026

Copy link
Copy Markdown
Contributor Author

Review round 1 (OpenAI gpt-5.6-sol, effort medium, 2026-10-23), reviewed SHA 33f11dd

Clean-room clone at the head, probes under .review/; CI green before the round. The report verbatim:

MEDIUM — spec/workflow.md:565, rfcs/0020-workflow.md:743: §9 orders admission checking before deterministic bound accounting, but never explicitly says admission status must be established before applying the later-sibling rule. A reader could incorrectly include a projection-failed instance. Fix: state in check 4 that §2’s later-sibling rule applies only after admission is checked and only to admitted instances.

Verified

  • §2 contains the admission qualification and §§3/7 cross-reference; §§3 and 7 are substantively unchanged.
  • Positive vector has the required four instances in order and expected states.
  • Independent probe confirms the positive vector and finds exactly the intended mismatch in the negative twin.
  • Negative twin differs only by projection-invalid → limit-exceeded; expected report is inventory-mismatch.
  • Changelog accurately records issue RFC 0020: a later sibling of a limit-exceeded instance whose projection failed (§2 against §§3, 7) #213.
  • tests.test_workflow: 18 passed.

merge: after fixes — 33f11dd

Disposition (the coordinator's)

# Finding Severity Disposition
1 §9's verifier checks never say that admission status is established before the later-sibling rule is applied MEDIUM Fixed in this pull request: one sentence in check 4 of §9, in the companion and in the RFC record; no further round

The merge follows CI on the fix. Same vendor as the builder, not a cross-vendor claim.

…dmission (#213 review)

Signed-off-by: kikashy <kikashy@hotmail.com>
@kikashy

kikashy commented Oct 22, 2026

Copy link
Copy Markdown
Contributor Author

The round-1 MEDIUM is fixed at 7cdba75 (check 4 of §9 states that the later-sibling rule is applied after admission is established and only to admitted instances, in the companion and the RFC record); CI green; no further round. Merged on the standing rule.

@kikashy
kikashy merged commit d09ccaf into main Oct 22, 2026
6 checks passed
@kikashy
kikashy deleted the rfc0020-213-209 branch October 22, 2026 19:14
@github-actions

Copy link
Copy Markdown
Contributor

🎉 Thank you for your contribution to Judgment Pack!

If you’d like another task, choose a path that interests you:

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.

RFC 0020: a later sibling of a limit-exceeded instance whose projection failed (§2 against §§3, 7)

1 participant