docs: record the adversarial re-review of PR #294 - #381
Closed
REPPL wants to merge 1 commit into
Closed
Conversation
An independent second pass over PR #294 and over this folder's first review, reproducing every claim by execution against a materialised merge of the PR head into main rather than reading the diff. The first pass missed the only red check on the pull request: the attribution gate fails on the PR body, which mentions the Assisted-by trailer inside prose where the check is line-anchored on purpose. Because the body step runs first and exits non-zero, the commit-trailer step never ran, so the summary's "attribution trailer correct" rested on a check that did not execute. Run locally, the commit half is clean; only the body is broken, and a body edit re-runs the gate with no new commit. Two substantive additions: the capture agreement test builds its map by hand and never parses text, so widening it cannot reach the parser-level disagreements it exists to catch -- a bare `impact:` decodes to a nested map and is refused as "must be a string" while record-lint calls it absent, and a single-quoted null survives unquoting where a double-quoted one does not; and the widening moves the acceptance surface at ten further IsNull call sites in both directions, turning a fail-closed spec_id error into a silent skip and a silent slug into a new blocker. Confirms F5, F7, F9, F10 by execution, verifies F3's scope claim (exactly two narrow copies existed, both widened, no import cycle in the proposed delegation), and verifies the fail-before/pass-after claim for all three new tests. Assisted-by: Claude:claude-opus-5
Deploying with
|
| Status | Name | Latest Commit | Updated (UTC) |
|---|---|---|---|
| ✅ Deployment successful! View logs |
abcd | 765e8cb | Aug 19 2026, 11:10 PM |
Collaborator
Author
|
Replaced: the branch's commit carries an AI git identity the attribution gate correctly refuses, and rewriting a pushed branch would need a force-push this repo forbids. Same content, correct identity: see the replacement PR. Assisted-by: Claude:claude-fable-5 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The independent second pass over PR #294 — commissioned adversarially over both the diff and the first-pass evidence already on main (
00-summary.md,01-findings.md, merged in #366). It lands02-adversarial-review.mdbeside them, completing the evidence trail.Why it matters: the second pass reproduced every claim by execution against a materialised merge and found the first pass missed the PR's one red required check (
attribution) and asserted the opposite — so the standing review's "happy to merge once the five points land" is unreachable as written; the author also owes the disclosure line. Two further findings sharpen point 4 and the iss-285/itd-128 deferral.Housekeeping: this branch sat unmerged from its session — nothing in this repo lands without a PR, and the session never opened one. Opening it now closes that gap; the reviews charter treats a new file beside an existing dated review as an append.
Assisted-by: Claude:claude-fable-5