chore(ech05): regression fixture → JCS, stage 6 first target, #431 row, owner norms [stack 3/3]

[hyperpolymath]

Summary

Stack 3 of 3. #432 and #433 have merged, so this PR now targets main. It holds the ECH-05 follow-ups the owner decided after #430 merged, plus the owner's norms written down where agent sessions read them.

History rebuilt, 2026-10-14 (owner decision: force-with-lease). The branch used to carry the pre-squash copies of #432 and #433. Rebase-merge replayed them onto main, which already had the squash commits, and conflicted on .hypatia-baseline.json. The branch is now this PR's own four commits on current main (5583c70), so it rebase-merges cleanly. Its tree is byte-identical to the old head (75288e7) merged with main.

Changes

• tests/fixtures/arbitration_regression.json → JCS (RFC 8785) I-JSON plus a final LF.
  • It is re-serialised, not regenerated. All 327 numbers, and the whole document, compare equal to the fixture recorded at 29b18ea; only spellings change (0.0 → 0). This is the source of the large deletion count.
  • The regenerator now writes JCS.
  • A new test, fixture_is_jcs_canonical, keeps it canonical. A pretty-printed fixture fails it; I ran that negative control.
• ROADMAP stage 6 and the gap-analysis pair table: HOL4 ↔ HOL Light through OpenTheory is the first target (owner decision, 2026-10-14). Lean 4 ↔ Rocq through Dedukti is the next candidate.
• issues/README.md: ECH-05 is ECH-05: feat(arbitration): consume fidelity evidence, not N re-proofs — type the arbiter's input by alignment certificate and level its verdict #431, closed as completed by feat(arbitration): ECH-05 fidelity evidence and belief/epi/sound verdict levels #430, and is the arbiter-typing slice of Consume fidelity evidence, not re-proofs #429, which stays open.
• .claude/CLAUDE.md, owner norms:
  • decisions are asked with the selector tool;
  • work lands as stacked PRs;
  • findings are fixed at source. The exemption ratchet admits ledger growth only with a Ratchet-exception: line naming the ledger.
  • The format standards point at their normative source in standards: YAML-POLICY.adoc (KYAML, .kyaml; D280/D307/D293), the workflow gate in standards#1025, JSON-POLICY.adoc (I-JSON as JCS; D306) and ADR-008 (UUID v8; D305/D305b/D320).
  • The UUID rule is scoped until the ids.rs migration lands. ids.rs still mints v7 records and v8 content ids without a domain. This is the wording the owner accepted on proof-burrower#111.

RSR Quality Checklist

Required

• Tests pass (cargo test --test arbitration_regression_test: 2 passed, 1 ignored)
• Code is formatted (cargo fmt --all -- --check)
• Linter is clean
• No banned language patterns
• No unsafe blocks
• No banned functions
• SPDX license headers present on all new/modified source files
• No secrets, credentials, or .env files included

As Applicable

• docs/ROADMAP.adoc updated (stage 6 first target)
• Documentation updated
• Others: n/a

Testing

$ cargo test --test arbitration_regression_test
test numeric_fusion_unchanged_and_level_is_belief ... ok
test fixture_is_jcs_canonical ... ok
negative control: pretty-printed fixture → fixture_is_jcs_canonical FAILED ("fixture must be JCS (RFC 8785) canonical")
jq: 327 numbers equal, documents equal (29b18ea fixture vs JCS fixture)

Stack

1. fix(deps): bump quinn-proto 0.11.15 → 0.11.19 (RUSTSEC-2026-0338..0342) [stack 1/3] #432: quinn-proto lockfile bump (merged)
2. chore(hypatia): delete dead ci-templates/mirror.yml, qualify echidnabot paths, baseline → JCS [stack 2/3] #433: Hypatia findings fixed at source, baseline → JCS (merged)
3. This PR: ECH-05 follow-ups and owner norms
4. fix(ci): read workflows as KYAML as well as block YAML (S4 prerequisites) #436: workflow checkers read KYAML (S4), stacked on this PR

🤖 Generated with Claude Code

https://claude.ai/code/session_019a95kk88TZ1iegsBK3kox3

[hyperpolymath]

@coderabbitai review

(Stacked on #433, so its base isn't the default branch and auto-review skipped it. The review should cover only cb6d2e3 and f643035.)

---

Generated by Claude Code

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 23 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: ASSERTIVE
• Plan: Advanced
• Run ID: 5b60efce-aff3-4825-9287-8577e9e4221b

📥 Commits

Reviewing files that changed from the base of the PR and between 5583c70 and d53c96b.

📒 Files selected for processing: 6

• .claude/CLAUDE.md
• docs/ROADMAP.adoc
• docs/design/ADVANCED_TYPING_GAP_ANALYSIS.adoc
• issues/README.md
• tests/arbitration_regression_test.rs
• tests/fixtures/arbitration_regression.json

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

⚠️ Action not completed

Review rate limited.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.