Public proof · redacted
Research Desk — 2026-09-10
Readable proof page derived from QC evidence. No secrets. Internal box paths are not the product proof — this page is.
ChatGPT was NOT used in the verified run. Synthesis runtime =
executor-synthesis.RESULT PASS · Run ID RD-2026-09-10-001 · Date 2026-09-10 JST
Checks
| Check | Outcome |
|---|---|
| Public Library SOURCES collected | PASS (8× HTTP 200) |
| Ollama structure/extract | PASS (botshelf-deep-research-3b, ~13.8s) |
| Proof Engine local audit | PASS |
| Honest runtime labels | PASS (executor-synthesis not ChatGPT) |
| Conflicts recorded | PASS (2) |
| Sequential envelopes | PASS |
| No X / no Verified flip | PASS |
Runtime table (sequential)
| Order | Runtime | Notes |
|---|---|---|
| 1 | web-fetch-curl | 8 public Library URLs HTTP 200 |
| 2 | ollama:botshelf-deep-research-3b | ~13.8s · llama3.2 3.2B |
| 3 | local-file-audit | Proof Engine markdown/JSON audit |
| 4 | executor-synthesis | Final brief · NOT ChatGPT |
Limitations
- Ollama soft invent risk on open Verified-criteria questions — monitored.
- Synthesis prefers HTML + Proof Engine evidence over model speculation.
- No Library Verified claim from this run.