PLATFORM MESH — PUBLISHED HISTORICAL LAB REPORT
Run date: 1 October 2026. Publication review: 14 October 2026.
Source: NEO8ra-GA #44, merged at 4c12cf08fe3c9f548e6ed848005e5ca5fde70ef2.
https://github.com/hashnet-colab/NEO8ra-GA/pull/44
The source repository requires owner-granted access; this summary is public.

The retained lab reports record 16 actual service-flow records plus 4,760
synthetic federation-workload records, 4,776 N8 FINAL confirmations, zero
pending records and rejection of an altered copy of each confirmation.
The five report SHA-256 values are published in evidence.json alongside this
file. On 14 October all five matched the retained reports, whose aggregate
counts and mutation outcomes matched the public summary. The current 18
adapter, receipt-proxy and soak regression tests passed.

This review reconciled retained reports and reviewed/tested the implementation.
It did not rerun every private rail proof or repeat the original service run.
The historical verifier uses separately pinned chain trust; this is not a
certification by an independent organisation. The new browser proof fixture is
synthetic and does not reverify these historical N8 FINAL confirmations.

Actual recorded stack: Platform Mesh charts 0.3.0, kcp 0.31.0, Private LLM 2.10.1,
Chat UI 0.10.6. Two recorded API requests returned HTTP 200. One agreed test
request took 1,003.85 ms against a 5,000 ms target at EUR 0.10/request.
The separate workload used 150 stable simulated identities over two accelerated
rounds, with 130 fan, 10 buyer and 10 seller roles per round.

Scope: finite single-operator laboratory run, synthetic organisations and test
prices. No invoice, payment or Chamber ruling. Resource observations do not
establish original audit actors or independent meter truth. Sustained hardware
acceptance is separate; the scheduled tests on iron start 15 October 2026.