[Arham-Tahir64]

A helper-star restoration can be invisible at D's target boundary yet leak arbitrary helper dirt into targets if it is delayed through A/scatter/B. This package checks that proposed boundary change on the pinned PR967 word and both PR1062 star words.

The exact condition is K(S-I)=0. Distinct leaf columns and nonzero actual center responses exclude every nonempty restoration subset: 2^221-1 cases for each 221-edge word and 2^236-1 for the candidate, by a written subset argument. Independent literal replay checks all 678 single-edge moves on all 661 formal columns in both directions; a separate Boolean interpreter checks every dirty leaf witness. Every witness changes at least eight targets while sources/helpers are eventually restored. The complete ADD count and paid profile stay unchanged.

The domain moves identical final D helper-restoration gates to the full-frame end of fixed C; compensation, continuation, operands and all other gates remain fixed. A nontrivial reader-kernel positive fixture passes all 16 inputs. Clean-only and all-ones fixtures show false acceptance; six mathematical control types and five on-disk corruption controls accompany the result. This is an exact scoped obstruction, not an unrestricted restoration theorem or a new kappa.

Validation on Windows x86_64 / CPython 3.12.14, standard library:

• python3.12 scripts/research_package.py --package research/r439-terminal-star-restoration: PASS, including six tests and isolated optimized export with all input bytes preserved.
• Metadata/source pins, selected-source arithmetic baseline and staged whitespace checks: PASS.
• Native multistage/whole-hosted replay, changed-word precision/tape/compiler admission, analytic transfer and Lean verification: unrun; no selected supplier adoption.

One new seven-file package contains the proof, full receipt, independent verifier, task/source metadata and explicit replay. Research began on current main c5108f0; the branch is rebased on ddb4371 with all seven pinned blobs and guidance unchanged. Contribution head: 2490dac.

The bit-helper-contract scope is explicitly changed because PR1194/1311 already cover restored semantic-D substitution. Final overlap covered 728 open PRs and 225 Discussion records with complete pagination, bounded by body/README/ledger searches. Discussion 1565 carries the claim and handoff.

Credit PR967/1062/915/926/1015/923 and the DreamingOfClouds/eumemic supplier lineage; woklerx's PR1017 owns the D omission atlas; mightyOil (Philipp)'s PR1203 and Sagar Thakrar's PR1199 own the general reader/stabilizer observation. Exact pending heads, attribution and open next question are in README/TASK. Arham Tahir, with substantial OpenAI Codex assistance; one agent, no independent human review. Apache-2.0.

[Mister-Meeseeks]

Batch review at 2490dac98eacc294bac29f136f0715343611d55e.

Merged to main, preserving the original contributor commits and package attribution.

Reviewed scope: Exact cut conjugation shows every nonempty omitted-restoration subset leaves a nonzero target defect in the three fixed words. Moving unchanged restoration gates to the end does not repair that defect.

The pinned scoped command passed; the batch ledger contains its command, input hashes, environment and log. This incorporates the research/tooling contribution and does not change the selected κ or discharge the inherited multiplication-theorem assumptions.

Batch review, attribution and evidence. Thank you for the contribution.