Read-only investigation at parent d2afe97e, VM bac6789. No compiler/runtime behavior changed.
Artifacts reused from the recorded compact root-range stage and frozen October 7 benchmark baseline.

python3 docs/_tools/analyze_linked_calls.py ARTIFACT...
PASS for 8 compact builtin/conformance artifacts and 4 frozen benchmark artifacts.
Exact commands and SHA-256 values are in provenance.json.
Container SHA and ranges, indexed-table lengths/endpoints, function value layouts,
block owner/index/instruction counts, framed call operands, imports/export signatures
and imported module semantic-hash references checked by the analyzer.

.toolchain/build/cargo/debug-paths-verify/debug/debug-paths-verify ARTIFACT...
PASS for all 12 recorded artifacts through public verify_artifact with default ArtifactLimits.
Verifier was source-built after root ranges #706; binary SHA-256 is in provenance.json.

Inventory consistency checks:
PASS for all 12 current artifact hashes; sum of per-module function counts equals total;
final-owner candidates do not exceed virtual sites; final-receiver candidates do not exceed dynamic sites;
per-model candidate function totals match module totals; owned payload totals match section sums.

Read-only analyzer negative checks on a temporary copy of boot:
PASS: flip one digest byte -> rejected with 'digest mismatch'.
PASS: set first CODE instruction length to zero and recompute container SHA-256
-> rejected with 'instruction frame bounds'. Originals remain unchanged.

Repeat runs with PYTHONHASHSEED=1/2 reproduce inventory.json and benchmark-inventory.json byte-for-byte.
All 8 compact artifact hashes match the recorded #706 archive; all 4 frozen benchmark hashes match
the original October 7 baseline. Current analyzer SHA-256 matches provenance.json.
No timed execution, compiler regeneration, JVM/native transport, Minecraft or release checks run.
Static estimates are not realized size savings, runtime frequencies or performance measurements.