Non-goals
Explicitly out of scope — do not re-add without revisiting the spec.
From spec §12:
Multi-journal / multi-node sync
Not solving edge-node ↔ home-journal reconciliation as a general problem. Remote modules can stream events to one central journal over Tailscale, but there is still only one journal.
Central schema registry service / formal schema evolution
A shared registry service or cross-module schema negotiation is out of scope.
Trove builds a local type catalog at startup instead: trove:// type URIs,
Trove Type Definitions (TTD) with RFC 8927 JTD payload contracts, blob-stored
schema bytes, and schema_ref on validated events. See
type catalog.
WASM guest runtime
Dynamic loading via go-plugin and gRPC is in scope; sandboxed guest execution is not.
Perkeep-style content model
No permanodes, claims, or GPG signing. The blob-store interface shape from Perkeep was kept; its object model was not.
General-purpose platform
Single-user tool built around specific sources — not "store anything for anyone."
Alternative journal backends
Postgres, MySQL, and other database backends are out of scope. SQLite is the
single journal backend per spec §2. Vector search stays in-file via sqlite-vec,
not a separate database.