diff --git a/FABRIC-3.md b/FABRIC-3.md
index 7919d6d..f987e51 100644
--- a/FABRIC-3.md
+++ b/FABRIC-3.md
@@ -1796,10 +1796,20 @@ pubkey-based model — flagged, not silently reused.
 mechanism as the word card, gated the same way by Zuse's eligibility list (§H.5) — see the
 `acl_ttl` field added above. Also resolved same day: the `flags` bit values above.
 
-**Not yet scoped (deferred within this node):** whether `blk_get_meta()`/`blk_set_meta()`
-need new FORTH word wrappers or stay C-only like `zuse_cert_seed`'s "no FORTH access"
-precedent; the actual repurposing edit to `block_subsystem.h`/`.c` itself — this pass (plus
-the `acl_ttl` pass before it) produced the full field design, not the code change.
+**Decided 2026-09-02: yes, FORTH wrappers.** Mirrors the existing word-level ACL split
+exactly — raw C accessors (e.g. `BLK-ACL-ALLOW@`/`!`, `BLK-ACL-TTL@`/`!`, `BLK-OWNER@`), actual
+elevation/policy logic composed in FORTH on top, presumably a new capsule alongside `ACL.4th`
+— per this project's own standing hard rule ("ACL policy belongs in `ACL.4th`, never in C...
+Compose in FORTH first"). `zuse_cert_seed`'s "no FORTH access" is a different case, not a
+counter-precedent: that field is private signing-key material, walled off from FORTH for
+key-material-specific reasons — block ACL fields (`owner_fp`, `acl_allow`, `acl_ttl`, `flags`
+bits) are ordinary ACL state, same category as `DictEntry`'s `acl_allow`/`acl_ttl` (which
+already have FORTH accessors), not key material.
+
+**Only remaining open item on this whole node**: the actual code edit to
+`block_subsystem.h`/`.c` plus the new FORTH-wrapper primitives and policy capsule — field/API
+design is now fully decided across this pass and the two before it (`acl_ttl`, `flags` bit
+values, this FORTH-wrapper decision), nothing implemented yet.
 
 ### F.5 — `WIREBIND` (breadcrumb only — followed into `RUNCAP` instead, 2026-08-27)
 
@@ -3688,8 +3698,13 @@ doubling as block state) uses orthogonal bits, not a mutually-exclusive enum. St
 bit 0 = `CLAIMED`, bit 1 = `MIGRATING` (serves `MIGSM`), bit 2 = `STALE` (serves `UNCLEAN`),
 remaining 61 bits reserved.
 
-`BMAPFMT`'s field design (TTL + flags bits) is now complete but **the code change was never
-made** — this refactor is presumably where it finally lands.
+**FORTH-wrapper question decided 2026-09-02: yes.** Mirrors the word-level ACL split — raw C
+accessors (`BLK-ACL-ALLOW@`/`!`, `BLK-ACL-TTL@`/`!`, `BLK-OWNER@`), policy composed in FORTH on
+top. `zuse_cert_seed`'s C-only precedent doesn't apply — that's key material, block ACL fields
+are ordinary ACL state like `DictEntry`'s.
+
+`BMAPFMT`'s field/API design (TTL + flags bits + FORTH wrappers) is now fully complete but
+**the code change was never made** — this refactor is presumably where it finally lands.
 
 ### H.7 — Messaging protocol groundwork (prerequisite to the message card)