fix(charter): §3 — clarify ordinary charter edits classification (canary #2 follow-up) #22

Merged

pdurlej merged 1 commit from claude/orders/order-f-2-followup-charter-edit-classification into main

2026-05-02 09:21:37 +02:00

claude commented

2026-05-02 03:17:17 +02:00

Collaborator

Summary

Addresses canary ζ #2 codex tech finding (medium risk) on PR #17 post-merge.

What canary #2 found

"Impact-size rule leaves some charter edits unclassified — Small excludes any charter touch, Medium only includes 'charter sub-rule add', and Large does not cover ordinary charter edits."

Codex was right. Under previous wording, a typo fix in charter had no tier mapping (excluded from Small, didn't match Medium "sub-rule add", not Large).

Fix

3 minimal wording adjustments to §3 review-by-impact:

Tier	Before	After
Small	"no charter touch"	"no charter sub-rule add or substantive prose change" + explicit "ordinary charter edits ≤30 LOC are Small" qualifier
Medium	"charter sub-rule add"	"OR substantive charter prose rewrite" added
Large	(no charter clause)	"OR cross-section charter restructure" added
Boundary note	(existed)	extended with charter-tier-mapping completeness clause

Net diff: +4 / -4 LOC. Same total length, more precise wording.

Scope (per amended §3)

This PR is Small: ≤30 LOC charter edit (4 lines changed), single-file, clarification of existing rule (not new sub-rule, not substantive policy change). Pre-authorized as PR #17 follow-up fix per operator's 2026-05-02 ~03:08 CEST message: "z powodu zakresu będziesz mógł sam wypchnąć".

This is also the first test of the amended §3 rule itself (recursive: amending §3 with §3 amended).

Other canary #2 dissents (NOT in this PR)

product-glm BLOCKER "Bootstrap paradox": moot post-merge (decision-log entry rather than code fix)
product-glm risk "small-PR self-classification can bypass safety net": already mitigated by ensemble re-classification clause in §3 ("may re-classify or reject with 'wrong size declared'")

Co-authors

codex tech reviewer (canary ζ #2 finding, 2026-05-02 ~03:09 CEST)
Claude Opus 4.7 (1M context)

## Summary Addresses canary ζ #2 codex tech finding (medium risk) on PR #17 post-merge. ## What canary #2 found > "Impact-size rule leaves some charter edits unclassified — Small excludes any charter touch, Medium only includes 'charter sub-rule add', and Large does not cover ordinary charter edits." Codex was right. Under previous wording, a typo fix in charter had no tier mapping (excluded from Small, didn't match Medium "sub-rule add", not Large). ## Fix 3 minimal wording adjustments to §3 review-by-impact: | Tier | Before | After | |---|---|---| | Small | "no charter touch" | "no charter sub-rule add or substantive prose change" + explicit "ordinary charter edits ≤30 LOC are Small" qualifier | | Medium | "charter sub-rule add" | "OR substantive charter prose rewrite" added | | Large | (no charter clause) | "OR cross-section charter restructure" added | | Boundary note | (existed) | extended with charter-tier-mapping completeness clause | Net diff: **+4 / -4 LOC**. Same total length, more precise wording. ## Scope (per amended §3) This PR is **Small**: ≤30 LOC charter edit (4 lines changed), single-file, **clarification of existing rule** (not new sub-rule, not substantive policy change). Pre-authorized as PR #17 follow-up fix per operator's 2026-05-02 ~03:08 CEST message: "z powodu zakresu będziesz mógł sam wypchnąć". This is also **the first test of the amended §3 rule itself** (recursive: amending §3 with §3 amended). ## Other canary #2 dissents (NOT in this PR) - product-glm BLOCKER "Bootstrap paradox": moot post-merge (decision-log entry rather than code fix) - product-glm risk "small-PR self-classification can bypass safety net": already mitigated by ensemble re-classification clause in §3 ("may re-classify or reject with 'wrong size declared'") ## Co-authors - codex tech reviewer (canary ζ #2 finding, 2026-05-02 ~03:09 CEST) - Claude Opus 4.7 (1M context)

claude added the

ready-for-operator

label

2026-05-02 03:17:17 +02:00

pdurlej was assigned by claude

2026-05-02 03:17:17 +02:00

claude added 1 commit

2026-05-02 03:17:17 +02:00

fix(charter): §3 — clarify ordinary charter edits classification (canary #2 follow-up) 47b1b7c1bd

Per canary ζ #2 codex tech finding (medium risk) on PR #17 post-merge:

> "Impact-size rule leaves some charter edits unclassified — Small excludes
> any charter touch, Medium only includes 'charter sub-rule add', and Large
> does not cover ordinary charter edits."

Codex correctly identified that ordinary charter edits (typo fixes, formatting,
clarifications, link repairs) had no tier mapping under previous wording.

Fix:

1. Small: changed "no charter touch" → "no charter sub-rule add or substantive
   prose change". Added explicit qualifier: ordinary charter edits ≤30 LOC are Small.

2. Medium: added "OR substantive charter prose rewrite" — distinguishes from
   sub-rule add but covers prose-level changes affecting policy interpretation.

3. Large: added "OR cross-section charter restructure" — for the rare large
   reorganization case.

4. Boundary determinism note extended: clarifies that the ordinary-charter-edit
   Small qualifier ensures every charter touch maps to exactly one tier.

Per charter §3 review-by-impact (NEW disjoint tiers): this PR itself is **Small**
— ≤30 LOC charter edit, single-file, clarification of existing rule (not new
sub-rule, not substantive policy change). Pre-authorized as PR #17 follow-up
fix per operator's 2026-05-02 ~03:08 CEST message.

Net diff: +3 / -3 LOC. Same total length, more precise wording.

Addresses 1 of 3 canary #2 dissent risks. Other 2 are non-actionable
(bootstrap paradox = moot post-merge; small-PR self-classification bypass =
already mitigated by ensemble re-classification clause in §3).

Co-authored-by: codex tech reviewer (canary ζ #2, finding)
Co-authored-by: Claude Opus 4.7 (1M context)

claude referenced this pull request from a commit

2026-05-02 03:19:21 +02:00

docs(status): STATUS_NOW night session refresh + §0b live ops artifacts index

claude referenced this pull request

2026-05-02 03:19:31 +02:00

docs(status): STATUS_NOW night refresh + §0b ops artifacts index #23

pdurlej commented

2026-05-02 09:09:56 +02:00

Owner

ok to merge @claude

i can’t do it from my mobile (i don’t know why)

ok to merge @claude i can’t do it from my mobile (i don’t know why)