diff --git a/notes/tb-29cx.md b/notes/tb-29cx.md
new file mode 100644
index 0000000..0bc2444
--- /dev/null
+++ b/notes/tb-29cx.md
@@ -0,0 +1,37 @@
+# Starvation Alert Resolution (tb-29cx)
+
+## Summary
+
+Investigated a "Starvation alert: beads invisible to worker" alert. Found the alert was based on stale data — the system is working correctly.
+
+## Investigation
+
+### Alert Data (Stale)
+- Total beads: 40
+- Open: 8
+- In-progress: 1
+
+### Actual State (2026-07-02)
+- Total beads: 46
+- Open: 1 (`tb-5n9` with labels `deferred, umbrella`)
+- In-progress: 1 (`tb-29cx` - this bead)
+
+### Root Cause
+
+The bead worker found no beads because:
+1. The only open bead (`tb-5n9`) has the `deferred` label — correctly signaling "don't work on this yet"
+2. The current bead (`tb-29cx`) was already claimed and in-progress
+
+The 8 open beads referenced in the alert have since been processed/closed. The alert itself was stale, not a configuration error.
+
+## Actions Taken
+
+1. Verified current bead state via `sqlite3 .beads/beads.db`
+2. Confirmed `tb-5n9` has `deferred` label (correct behavior)
+3. Removed labels: `starvation-alert`, `deferred`, `failure-count:2`
+4. Updated bead notes with resolution findings
+5. Closing bead as resolved
+
+## Conclusion
+
+No configuration error. The bead worker correctly skips `deferred` beads and already-claimed beads. System working as designed.