[AdaWorldAPI]

CC1's first falsifier, claimed on EXO-50 (session_01VncKoyG6JkpsoqhqTGTv3f/036) with the ownership sync GPT:gpt-ada-01/20261023-cc1-cc2-ownership-sync-13. It is lab only. There are no production edits and no contract changes.

Question

A squash replaces a family of lane chains with one delta. For which readers does the squash give the same answer as the full history, and which would be lost? When a protected reader cannot be answered, the squash must keep a witness or be refused. A refusal is an expected outcome here, not a failed test.

The probe was pre-registered in entries/2026-10-23-d-tga-5d-protected-readers-preregistration.md (commit fb2e317) before any code was written.

Changes

• src/squash.rs: the D-TGA-5 gate moved out of tests/squash_gate.rs. Only visibility changed; all 10 existing tests are unchanged and pass.
• protected_squash(givens, prior, lanes, protect): runs R1/R2 unchanged, then answers each protected reader from the retained data alone. If a reader cannot be answered, it refuses with Unretainable(reader). The readers are:
  • Final
  • Intermediate
  • Contradictions
  • Divergences
  • Multiplicity
• tests/protected_readers.rs: 7 tests over fixtures (a)–(e):
  • (a) pruned, vetoed and planned writes;
  • (b) a contradiction followed by a rewrite;
  • (c) an intermediate divergence followed by agreement;
  • (d) a cell written 3× by one lane and 1× by another;
  • (e) a non-empty prior version.
• STATUS_BOARD gains a D-TGA-5D row; D-TGA-5 is marked Shipped (lab, probe(D-TGA-5): squash the chain of origin into the most-updated lane, gated on contradiction; blame survives #1736). The probe was first called D-TGA-6, but that id is MergeQueue, so it was renamed.

Results

• Final reads are kept on every fixture with an empty witness.

• Without a witness, the squash loses (b)'s contradiction and (c)'s divergence. On (d) it reports multiplicity 2, because it counts lanes; the history has 4 writes.

• With the logs protected, the witness keeps all three and every answer matches the history.

• A protected intermediate reader refuses the squash exactly when an intermediate read differs from the final read. That is 3 of 5 fixtures; the silent twins are accepted.

• Bytes (reported, not gated). One lane, 8 cells:

  k	transient	squash	witness
  1	104	221	72
  4	416	236	72
  16	1664	296	72
  64	6656	536	72

  At k = 1 the squash is larger than the history it replaces.

Limit: for the three logs, the witness is built from the same history it is checked against, so this check only catches wiring bugs. What the probe measures is what a bare squash loses, what keeping it costs, and when the squash must be refused.

Disables

Each disable was run after the commit, with its anchor asserted:

disable	tests that turn red
squash landed on an empty version	d1, d3
contradictions not filled	d3
divergences not filled	d3
multiplicity not filled	d3, d5
intermediate check removed	d4

Gates

In crates/thought-git-probe: cargo fmt --check, cargo clippy --all-targets -D warnings and cargo test (all 8 test binaries green). entries_index.py --check is clean, and SUPERSESSION-INDEX.md is current.

Not in this PR

• F5 (synthesized head)
• F6 (correlated derivations)
• the passive ProofRef: the type still has to be agreed with CC2 before any contract PR

---

Generated by Claude Code

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 67a77e61-859c-4b75-8e9d-13529469d14b

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

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

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.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-23T10:55:09.109765Z	52fab2c	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.