docs: record public release verification and storage follow-up

docs/release-1.9.0-acceptance.md

@@ -978,3 +978,35 @@ and companion0.5.34 APK all passed full download byte-count and SHA256 checks:
 locally verified pinned-root signatures. Promote those exact bytes to the live
 metadata paths; no artifact or signature changed. ISO upload and demo deployment
 are separate ongoing operations, not inferred passed from OTA asset publication.
+
+### Publication status and subsequent storage regression
+
+OTA metadata promotion proposal
+`0e90b3847ff377ad5c9e400037651b565defa4ee229eac785a7897c30423a4e7`
+was marked applied. Exact main/tag parity and public metadata bytes passed. Dev,
+Yaya and Shorty now use the public catalog; their app IDs/start times and guard
+configuration were preserved. Thirty-two external management-denial probes and
+tailnet UI access passed after the change. Framework reaches the public manifest
+and its saved current version is already1.9.0-alpha; this is a read-only result,
+not a fresh authenticated update RPC acceptance.
+
+The public demo was deployed with immutable qualified images and preserved
+configuration/rollback. Public version is1.9.0-alpha-demo. Mobile login/dashboard
+passed on retry after the initial30-second form wait timed out. Public resumable
+upload recovery/exact bytes/visitor isolation and replace/zero-byte/cancel cases
+passed. Fixed-quota and interrupted-TCP scenarios retain their isolated evidence.
+
+The raw ISO and both checksum files passed full public-download hash checks.
+The public signed checksum also verifies against the pinned release root. Logs:
+`/tmp/archy-190-publish-iso.log` (ISO PASS before the idle upload tunnel closed),
+`/tmp/archy-190-publish-iso-checksums.log` (small sidecars retried over HTTPS).
+No ISO re-upload or artifact change was needed. A subsequent
+follow-up backend suite exposed the pre-existing storage-prefix bug documented in
+[known limitations](release-1.9.0-known-limitations.md). Its correction is being
+qualified separately. Artifact-byte verification is not a claim that this newly
+identified issue has been fixed in the already published release.
+
+Public assets: [1.9.0-alpha release](https://source.archipelago-foundation.org/lfg2025/archy/releases/tag/v1.9.0-alpha).
+The separate follow-up branch now passes1,683 backend tests (four existing ignored)
+and all1,222 frontend tests after correcting the storage classifier. These are
+source test results, not inclusion in the published artifacts or live acceptance.

docs/release-1.9.0-known-limitations.md

@@ -1,4 +1,6 @@
-# 1.9.0-alpha: Angor historical discovery
+# 1.9.0-alpha: known limitations
+
+## Angor historical discovery
 
 Historical project discovery is incomplete. Funding commitments for all35
 reference projects were verified, but34 original signed announcements were not
@@ -16,3 +18,20 @@ The dev node is still syncing; full-chain evidence comes from Shorty.
 Evidence and inventory: [client acceptance](angor-client-acceptance-20261001.md),
 [project recovery inventory](angor-project-recovery-20261001.json), and
 [release acceptance](release-1.9.0-acceptance.md).
+
+## Chat/contact storage migration — discovered during follow-up testing
+
+A subsequent full backend test run exposed an existing random-prefix collision
+in `storage_crypto::is_plaintext_json`: an encrypted store whose nonce begins
+with `{` or `[` can be mistaken for legacy plaintext. This can prevent chat or
+contact state loading correctly, and the message migration path can overwrite
+that state. This helper is used for chat messages and mesh contact customizations,
+not wallet databases.
+
+The follow-up branch replaces the prefix-only heuristic with complete JSON
+validation and adds deterministic encrypted-prefix regression cases. Qualification
+is in progress; the published 1.9.0-alpha artifacts do not include that fix.
+Existing release signatures and tags must not be silently replaced. Track a
+corrective update and preserve affected state before further node restarts.
+The earlier green release run did not detect this probabilistic failure; do not
+interpret it as proof that this newly discovered issue is absent.