Skip to content

v0.25.0 - Runtime schema packs foundation

Choose a tag to compare

View all tags
@solaitken solaitken released this 30 May 15:26
· 151 commits to main since this release
v0.25.0
f62918c
This commit was created on GitHub.com and signed with GitHub鈥檚 verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Open Second Brain v0.25.0 - Runtime schema packs foundation

Open Second Brain gains a runtime schema-packs foundation: a vault can now declare custom taxonomy tokens in Brain/_brain.yaml, artifacts can carry optional inert schema_type: metadata, and operators can inspect the resolved vocabulary and live usage through a read-only o2b brain schema report - all while plain Markdown and YAML stay the source of truth, validation stays inside one vocabulary boundary, and the deterministic dream path is unchanged.

Open Second Brain 0.25.0 - a runtime schema vocabulary lets _brain.yaml declare preference_types, signal_types, page_types, and log_event_kinds behind one validation boundary with forward-compatible warnings, artifacts carry optional inert schema_type metadata that leaves dream, retention, search, and lifecycle enums untouched and keeps legacy files byte-identical, and a read-only o2b brain schema report surfaces resolved vocabulary, usage counts, unknown tokens, and unused declarations while mutation primitives, MCP admin ops, CLI verbs, and a schema-author skill stay deferred behind the ADR

What ships

  • Runtime-extensible taxonomy. Brain/_brain.yaml schema: can declare preference_types, signal_types, page_types, and log_event_kinds on top of the built-in tokens, with no source changes or migrations.
  • One validation boundary. src/core/brain/schema-vocab.ts owns normalization, token-shape validation, deterministic merge/dedupe, and known-token lookup, so parser and writer share one vocabulary.
  • Inert schema_type metadata. Preferences, retired preferences, and signals can carry optional schema_type: metadata; writers omit the field unless supplied, so legacy artifacts stay byte-identical. It is taxonomy metadata only - dream, retention, search ranking, lifecycle policy, and page-state enums are untouched.
  • Forward-compatible config. Unknown schema: subkeys warn rather than fail, matching the existing config-evolution style.
  • Read-only inspection. o2b brain schema [--json] reports the resolved vocabulary, artifact token-usage counts, unknown tokens, and unused custom declarations without writing to the vault.

Notes

  • This release ships the foundation only. Full schema mutation primitives, MCP admin operations (e.g. schema_apply_mutations), the broader CLI verb surface, and a schema-author skill stay deferred behind the foundation ADR in docs/brainstorm/runtime-schema-packs-foundation/adr.md.
  • schema_type is taxonomy metadata only in this release and performs no behavior changes; the schema report is read-only and writes nothing.
  • Full suite green on merge: 2872 tests passing, typecheck and lint clean, canonical version 0.25.0 synced across all targets.
Assets 4
Loading