diff --git a/.huskies/work/1_backlog/490_story_crdt_state_layer_backed_by_sqlite.md b/.huskies/work/1_backlog/490_story_crdt_state_layer_backed_by_sqlite.md
new file mode 100644
index 00000000..e8de2013
--- /dev/null
+++ b/.huskies/work/1_backlog/490_story_crdt_state_layer_backed_by_sqlite.md
@@ -0,0 +1,24 @@
+---
+name: "CRDT state layer backed by SQLite"
+---
+
+# Story 490: CRDT state layer backed by SQLite
+
+## User Story
+
+As a developer, I want the BFT JSON CRDT document backed by SQLite for persistence, so CRDT ops survive restarts and the state layer is ready for multi-node sync.
+
+## Acceptance Criteria
+
+- [ ] BFT CRDT crate (crates/bft-json-crdt/) integrated into the server
+- [ ] CRDT ops persisted to SQLite via sqlx (backend-agnostic schema)
+- [ ] Pipeline state reads switch from filesystem to CRDT document
+- [ ] Pipeline state writes go through CRDT ops (which persist to SQLite)
+- [ ] Filesystem .huskies/work/ directories still updated as a secondary output for backwards compat during transition
+- [ ] CRDT state reconstructed from SQLite on startup
+- [ ] agent: coder-opus
+- [ ] depends_on: [489]
+
+## Out of Scope
+
+- TBD