DungeonQ 0.14.0: inspect a lost write reply before retrying

Following Theodor’s question on the previous walkthrough, I added a concrete interrupted-write case and a read-only recovery path.

What changed:

• The operator can distinguish a verified canonical commit, an authenticated no-effect refusal, an unknown outcome and a conflicting request identity.

• The participant and standalone MCP consumer can recover an already-saved result without sending another write.

• The Control room marks committed writes beside their route evidence. Unknown attempts remain visible.

• A new 72-second English walkthrough, raw result JSON and source snapshot are available from one page.

The recorded test cuts the reply after a durable note save, recovers the original result, checks a separate readback and restarts the gateway. It records one write and zero automatic write retries. Refusal and before-execution loss are tested separately.

The source passed 500 tests plus public Linux/macOS and source-bound runtime/container checks. The deliberately faulted demonstration itself still reports failed transport evidence; recovery does not rewrite history.

Watch, inspect or reproduce: https://dungeonq-astra.kq7dn7jb6r.chatgpt.site/#write-recovery

This is a local artificial-resource engineering test. The October 8 host/model film remains historical; no new model run or production-protection claim is included.