Register semantic retention evidence in source readers

This commit is contained in:
Meta Vens 2026-08-10 04:06:01 +01:00
commit 7e320b5d14
5 changed files with 77 additions and 3 deletions

View file

@ -2664,3 +2664,14 @@ ready-chain preservation, and rejection of hostile future generations even
inside the recent window. A warning-free production binary check covers the
daemon wiring. Exact packaged copied-live proof remains the acceptance boundary;
physical Qdrant epoch cleanup remains a distinct executor-owned action.
The first exact packaged copied-state run exposed a catalog crossing omitted by
the initial map. Logical retention writes into the canonical Mind and Modeling
stores, so every general reader of those stores must recognize the evidence-head
schema even though only the retention owner may write or interpret it. The
projector cache already registered the head; `agent_memory_cache` and
`runtime_spine_cache` did not, causing current semantic input/health reads to
refuse the entire store after the first successful retirement. Both canonical
source catalogs now register the head as evidence-only. Projection-input tests
persist a head and prove current Mind and Modeling derivation remains unchanged.
This is schema legibility, not shared decision authority.