diff --git a/engram/dist/engram b/engram/dist/engram
index f3178b7..403fd55 100755
Binary files a/engram/dist/engram and b/engram/dist/engram differ
diff --git a/lang/releases/v1.0.0-20260501/el_runtime.c b/lang/releases/v1.0.0-20260501/el_runtime.c
index 2314047..0445841 100644
--- a/lang/releases/v1.0.0-20260501/el_runtime.c
+++ b/lang/releases/v1.0.0-20260501/el_runtime.c
@@ -5578,11 +5578,24 @@ void el_cgi_init(el_val_t name, el_val_t dharma_id, el_val_t principal,
  *   when a suppressed node breaks through.
  * ENGRAM_INHIBITION_FACTOR: multiplier applied to working_memory_weight when
  *   an inhibitory edge fires against a node (0 = full suppress, 0.3 = partial). */
-#define ENGRAM_WM_THRESHOLD          0.15
+/* ENGRAM_WM_THRESHOLD lowered 0.15 → 0.08 (2026-06-04 self-review):
+ * With 29K nodes and only ~100 edges the graph is extremely sparse. BFS
+ * fan-out carries very little activation energy — most background values
+ * arrive at 0.05-0.12 through the thin path set. The prior 0.15 threshold
+ * was gatekeeping too aggressively, suppressing legitimate weak signals.
+ * Cognitive literature (TBRS*, Altmann/Schunn) uses θ=0.05; 0.08 is a
+ * conservative move in that direction that still filters random noise. */
+#define ENGRAM_WM_THRESHOLD          0.08
 #define ENGRAM_WM_DECAY              0.7
 #define ENGRAM_SUPPRESSION_BREAKTHROUGH 5
 #define ENGRAM_BREAKTHROUGH_WEIGHT   0.25
-#define ENGRAM_INHIBITION_FACTOR     0.1
+/* ENGRAM_INHIBITION_FACTOR raised 0.1 → 0.2 (2026-06-04 self-review):
+ * Factor=0.1 produces 90% suppression per inhibitory edge hit. On a sparse
+ * graph where edges are rare, this nearly always silences the target node
+ * entirely. Factor=0.2 (80% suppression) retains meaningful inhibition
+ * while allowing partially-suppressed nodes to remain faintly active,
+ * consistent with partial inhibition models in cognitive neuroscience. */
+#define ENGRAM_INHIBITION_FACTOR     0.2
 
 /* ── Layered consciousness architecture ──────────────────────────────────────
  *
@@ -8917,6 +8930,24 @@ el_val_t engram_wm_count(void) {
     return (el_val_t)count;
 }
 
+/* Average working_memory_weight across all promoted nodes (wm > 0).
+ * Returns the float bit-pattern via el_from_float so EL can use it with
+ * float_to_str / float_gt. Returns 0.0 when no nodes are promoted.
+ * Useful in heartbeat ISEs to distinguish "many weak activations" (sparse
+ * graph, low avg) from "few strong activations" (dense subgraph, high avg).
+ * Added 2026-06-04 self-review for graph health observability. */
+el_val_t engram_wm_avg_weight(void) {
+    EngramStore* g = engram_get();
+    double sum = 0.0;
+    int64_t count = 0;
+    for (int64_t i = 0; i < g->node_count; i++) {
+        double w = g->nodes[i].working_memory_weight;
+        if (w > 0.0) { sum += w; count++; }
+    }
+    double avg = (count > 0) ? (sum / (double)count) : 0.0;
+    return el_from_float(avg);
+}
+
 /* engram_list_layers_json — serialized counterpart of engram_list_layers.
  * Returns a JSON array, sorted by activation_priority ascending. */
 el_val_t engram_list_layers_json(void) {
diff --git a/lang/releases/v1.0.0-20260501/el_runtime.h b/lang/releases/v1.0.0-20260501/el_runtime.h
index 8a22b73..c733dd6 100644
--- a/lang/releases/v1.0.0-20260501/el_runtime.h
+++ b/lang/releases/v1.0.0-20260501/el_runtime.h
@@ -619,6 +619,7 @@ el_val_t  engram_neighbors_json(el_val_t node_id, el_val_t max_depth, el_val_t d
 el_val_t  engram_activate_json(el_val_t query, el_val_t depth);
 el_val_t  engram_stats_json(void);
 el_val_t  engram_wm_count(void);
+el_val_t  engram_wm_avg_weight(void);  /* avg wm weight of promoted nodes; float bits */
 el_val_t  engram_apply_decay_json(void);
 el_val_t  engram_list_layers_json(void);
 /* engram_compile_layered_json — produce a prompt-ready text block split