# Fresh Publishing Desk five-role result

Run date: September 20, 2026
Tested bundle SHA-256: `5935128a84df986bdc891643b6e0ba45400ad6b6b0ad2bea34d26cbb4e857d97`
Locked source SHA-256: `ba8faf8619afebe2077c570608565ea1f94cf34e967edbd3d1560dcf10ed4cca`
Tools: none
Human approval: not given

Scout, Researcher and Story completed fresh runs from the tested bundle. The initial Writer contract duplicated Editor's exhaustive audit, produced oversized responses and failed length acceptance. A separated-responsibility Writer contract then passed after one measured length correction. Editor audited the exact draft boundaries using 24 immutable evidence options.

The preserved Editor response passed the authoritative structure-and-evidence gate with zero issues. Its original schema prefilter rejected populated legacy citation fields even though they agreed with the evidence arrays. Reconciliation of that schema with the gate's existing consistency rule validated the same raw response without changing it or making another model call.

Technical result: **the fresh five-role handoff completed and passed its structure-and-evidence gate**.

Semantic result: **publication held**. Codex inspection found one overclaim: “There is no commercial sponsor” is stronger than the source's “The report identifies no commercial sponsor.” The remaining claims were direct quotations or meaning-preserving paraphrases. This is not human approval or proof of general crew reliability.

Provider-derived estimates across successful stages and rejected correction attempts totaled `$0.012231882`. This is an estimate, not an invoice. No profile or shared authentication copies remain in the evidence directory.