Retry of the verified Atlas release after a snapshot failure

Railway deployment 4c8215bb-653e-45ad-9862-7e26336e0ac0, for commit 59dbb64ceaf83e8775ba54f8f24457ae10c0b844, failed before producing an image. The available build logs contain source snapshot uploads and archive unpacking, with no application build output. Railway's automated diagnosis classifies the failure as infrastructure related. The precise lower-level cause is not visible in the available logs; repository size is a possible factor mentioned by the diagnosis, not an established cause.

The previous verified deployment, b022d3b2-3b5f-496e-b243-447c3e645f24 at commit 416f67452b6b5dc7ef8bd6db1ec345feba227991, remained online with one healthy replica. The existing 50,000 MB /data volume remained mounted. The connected tools reported that the failed attempt had no reusable build (canRedeploy=false). Restarting the healthy release or creating another service would not recover the missing image.

The recovery publishes these compact failure records and operating instructions through the existing GitHub branch/CI/main workflow, requesting a fresh automatic build. It changes no application code, artwork, asset data, source permission, service configuration, volume or save. The eleven original-source findings, thirteen actual contexts, local Atlas validation and every remaining qualification gate stay intact. The first failed attempt remains recorded separately from the retry; neither is described as live until its actual deployment and live Atlas checks pass.

The user's standing routine-release authorization in AGENTS.md covers this existing repository and production service. No additional account login, manual acceptance step, fixed commit pin or destructive infrastructure operation is requested.

Recovery scope recovery-review.json. Build and runtime logs build-runtime-logs.json. Railway's automated diagnosis railway-diagnosis.json. Observed health environment-health.json. Service configuration service.json. Asset qualification.