Fresh census integration onto main, 2026-10-27

Base: be0eb25fc33f3b057847d6c4e2c4fe5a6a9a426c. Accepted production candidate: 13f1c27963df0037bba47acf035faec77f53898b. Independent acceptance revision: 234860ed740806e792dc1e063a1be55ff2273dad.

Only tools/build.sh, tools/check-region-decls.js, tools/region-census.js and tools/region-layout.js receive production changes. Their accepted blobs and the 29 enumerated tests, fixtures and reports are copied exactly from the independent acceptance revision. No candidate ancestry is merged. Main's compiler, DirectInput, sources, generated mirrors and ratchet baselines remain unchanged. Build ordering remains census --gate, region-map --check, then --build-gates running js-copies, embedded-wat, dup-payloads, hand-rolled.

The compiler context differs from the donor in tools/watx.js and tools/watx-src/compiler.js. The unchanged portable helper stages the actual main compiler files for both original and candidate processes; no compiler feature or dependency was imported. Full-file identities and the complete production JavaScript scan closure are in the integration receipt's archive.

Merged-tree validation:

• First behavioral gate: unchanged production-first.cjs, including the invalid align-3 declaration appended after reading host.js. Separate original CLI processes and the production combined CLI both reject at embedded-wat.
• Unchanged freshness differential driver: 24 cases, 90 CLI invocations, covering declaration and law-provider discovery, manifest changes, JS/WAT changes, same-size/restored timestamps and failure ordering.
• CLI contract: 147 production CLI invocations pass, including legacy modes, guard ordering, unsupported combinations, recording and API isolation.
• Disabled layout, declaration and manifest freshness each cause the unchanged corresponding differential oracle to reject. Causal diagnostics are retained.
• Real main-tree census --gate, generated mirror freshness, combined four gates and declaration validation pass. All 2,223 main JavaScript test paths (20,281,228 bytes) are present for recursive production scans. The embedded scan checks 1,033 fragments. Main baselines are unchanged.
• Tier discovery passes for 2,159 tests; the new test is discovered by the existing unit convention. Timeout validation and whitespace checks pass.

Reuse the independently accepted evidence by exact identity:

• scratch/usage-watch/airstrike-census-freshness-independent-accept-oct27/receipt.json SHA-256 2cbab6969c25a6787538ed9867dfe3d464f84cb2789e907cfa21547faa1495ea.
• Its durable archive SHA-256 758d01f3fb51eb4e4fe50b58a593d7a5df34dbfe45711c57ab57bc452c014408.
• Parent repair receipt SHA-256 914261c02793047daccddc2ea34dd8f8290d78e36b7ef5cf7a59eecedbc22332.

The canonical new delta, command receipts, compiler/source/test pins, controls, resource admissions, incidents and exact publication identities are retained in scratch/usage-watch/airstrike-accepted-fresh-census-main-integration-oct27/receipt.json and its evidence archive. That receipt is authoritative for publication state.

This qualifies source and CLI behavior only. No full build, canonical module compile, guest, browser, native or Boat execution occurred. Historical CPU rejections 126.1068 > 90 and 105.48018 remain binding; retained modules remain UNQUALIFIED and whole-invocation CPU completeness UNKNOWN. Tiny fixture counts (reads 10 to 5, parses 2 to 1, compiles 3 to 1) establish no full-build savings.