Original Everon now includes extraction in random clear-ground placement. Its former airport coordinate remains only a checked fallback, not the first choice. Automatic terrain placement samples world bounds from the first candidate and no longer starts with an airport discovery/preference pass. Authored scenarios can still explicitly require an airfield for their own fixed pads.

Primary dry-ground and aircraft/overhead clearance, guard footing, rescue-fleet space and the prison-relative escape boundary remain required. Terrain-independent placement also retains its streamed-navigation checks. Original Everon now checks the primary aircraft footprint before committing a random candidate and checks extraction-to-prison escape range in either placement order. The final site-spawn check applies to randomized original-Everon pads as well as terrain-generated ones. Scenario validation now accepts extraction's RandomClear flag.

Verification

The full python3 tools/check.py run passed, with output in repository-checks.log. The final validation-rule change and boundary tests also passed separately in final-generated-site-checks.log. The focused terrain scheduler and maintenance stock checks passed. Generated-menu freshness and git diff --check passed.

The native fixture uses frozen production addon sources and a private Workbench profile. It keeps the live scenario definitions and normal generation instead of switching to the historical fixed-site catalog. --terrain enables the actual terrain-independent placement path on Everon's native world/navigation. Neither mode bypasses ground, clearance or navigation checks.

Original-Everon run E3 completed normal preparation and admitted a player. It selected an extraction point approximately 5,885 m from the former airport pad and 684 m from the prison, without falling back to authored coordinates. The fixture verified the resolved non-airfield definition, escape-boundary distance, dry ground and rescue-fleet space. Evidence is in original-everon/.

Terrain-independent run E4 also completed normal preparation and player admission. It selected a landing zone approximately 6,665 m from the old airport pad and 1,399 m from the prison. Evidence is in terrain-independent/. Both final runs report smoke-completed, no script/runtime failures, confirmed private-profile cleanup, and unchanged production sources and installed dependencies.

The first diagnostic run E1 exceeded the native compiler's expression limit in a test-only fixture exclusion. E2 exposed the production definition validator's old rejection of randomized extraction. Both were corrected before E3. Their private test profiles were cleaned; the failed diagnostics were not treated as successful verification.

These are scripted local Workbench checks, not dedicated multiplayer or a complete rescue-flight/landing test. The strict native report still lists installed localization resource and world-property errors, so its process exit code is nonzero even when smoke assertions complete. Consult the raw runner report for those errors and the separate cleanup/source-integrity results.

Reproduce with fresh alphanumeric suffixes:

python3 docs/evidence/extraction-anywhere-2026-10-28/isolated-runner.py E5
python3 docs/evidence/extraction-anywhere-2026-10-28/isolated-runner.py E6 --terrain