Completed the full link-delivered LXMF unit:

Tier 1 audit: `link-lxmf-tier1-rns-1.2.4-lxmf-0.9.7.md`
Tier 2 vectors/verifier: link-lxmf.json, regen_link_lxmf.py, and verify_link_lxmf.py
Tier 3 promotion: updated SPEC.md, flows, status, and documentation
Key correction: the 319/320 boundary uses upstream’s computed LXMF content_size, not simply raw message content length.
Also corrected stale flow descriptions for KEEPALIVE (0xFA) and encrypted LINKCLOSE teardown (0xFC).
Verification:
Deterministic vector regeneration: identical SHA-256
Portable-path and formatting checks: pass
Full pinned suite: 17 passed, 0 failed

SPEC.md

@@ -686,6 +686,16 @@ destination_hash(16) || source_hash(16) || signature(64) || msgpack_payload(...)
 
 Full layout. The Link's session key encrypts the whole blob.
 
+For DIRECT delivery, upstream computes
+`content_size = len(msgpack_payload) - TIMESTAMP_SIZE - STRUCT_OVERHEAD` and
+selects a single Link DATA packet when `content_size <=
+LINK_PACKET_MAX_CONTENT`; otherwise it sends the complete body as a Resource.
+With default RNS 1.2.4 / LXMF 0.9.7 parameters the boundary is 319/320. The
+threshold applies to this computed LXMF content size, not simply the raw
+`content` field and not the complete signed body. At the 319 boundary, the
+complete canonical body is 431 bytes (`Link.MDU`) and the Link-encrypted wire
+packet is 499 bytes. (verified by `tools/verify_link_lxmf.py`)
+
 ### 5.3 `msgpack_payload`
 
 A msgpack array of 4 elements (5th optional):
@@ -2411,7 +2421,7 @@ The repeater repo's `pre_build.py` patches several other microReticulum protocol
 
 ## 10. Resource fragmentation protocol
 
-A **Resource** is the standard RNS mechanism for transferring a payload that exceeds the per-packet content limit of an established Reticulum Link. LXMF, Link REQUEST/RESPONSE, NomadNet, and file-transfer utilities use it for large bodies; an application can also define its own sequencing protocol or use Channel, so Resource is not the only possible large-data mechanism. With default RNS 1.2.4 / LXMF 0.9.7 parameters, DIRECT LXMF selects Resource when content exceeds **319 bytes** (`LINK_PACKET_MAX_CONTENT = Link.MDU(431) - LXMF_OVERHEAD(112)`, verified by `tools/verify_resource.py`). Resource is built **on top of** an active Link — it relies on the Link's session key for encryption (§3.1 link-derived form) and on the Link's bidirectional DATA channel for control traffic.
+A **Resource** is the standard RNS mechanism for transferring a payload that exceeds the per-packet content limit of an established Reticulum Link. LXMF, Link REQUEST/RESPONSE, NomadNet, and file-transfer utilities use it for large bodies; an application can also define its own sequencing protocol or use Channel, so Resource is not the only possible large-data mechanism. With default RNS 1.2.4 / LXMF 0.9.7 parameters, DIRECT LXMF selects Resource when its **computed LXMF content size** exceeds **319 bytes** (`LINK_PACKET_MAX_CONTENT = Link.MDU(431) - LXMF_OVERHEAD(112)`, verified by `tools/verify_link_lxmf.py`). This is `len(msgpack_payload) - TIMESTAMP_SIZE - STRUCT_OVERHEAD`, not necessarily `len(content)` and not the complete signed body. Resource is built **on top of** an active Link — it relies on the Link's session key for encryption (§3.1 link-derived form) and on the Link's bidirectional DATA channel for control traffic.
 
 The complete reference is `RNS/Resource.py` (1380 lines in RNS 1.2.4); `RNS/Packet.py:74-79` defines the context constants. This section describes the wire-level invariants a clean-room implementation must respect; many implementation choices (window sizing heuristics, watchdog timers, EIFR computation) are private and listed only when their absence would cause an interop break.
 
@@ -2419,7 +2429,7 @@ The complete reference is `RNS/Resource.py` (1380 lines in RNS 1.2.4); `RNS/Pack
 
 Three triggers in upstream:
 
-1. **`LXMessage.send()` for `DIRECT` method with `representation == RESOURCE`.** Set automatically when the encrypted-form LXMF body exceeds `LINK_PACKET_MAX_CONTENT` (`LXMF/LXMessage.py:415-421`).
+1. **`LXMessage.send()` for `DIRECT` method with `representation == RESOURCE`.** Set automatically when the computed LXMF `content_size` exceeds `LINK_PACKET_MAX_CONTENT` (`LXMF/LXMessage.py:405-421`; verified by `tools/verify_link_lxmf.py`).
 2. **NomadNet page request fulfillment** — a server returning a page whose body exceeds the link MTU.
 3. **Direct file transfers** via `rncp` and similar utilities.
 
@@ -3770,8 +3780,9 @@ See [`test-vectors/`](test-vectors/). Currently populated:
 - **`lxmf.json`** — deterministic opportunistic LXMF plaintext and Token ciphertext vectors. Verified by `tools/verify_lxmf_opportunistic.py`; regenerated by `tools/regen_lxmf.py`. Covers SPEC.md §3 and §5.
 - **`links.json`** — Link handshake and LRRTT vectors, including LINKREQUEST, LRPROOF, derived keys, and the activation packet. Verified by `tools/verify_link_handshake.py` and `tools/verify_link_lrrtt.py`; regenerated by `tools/regen_links.py`. Covers SPEC.md §6.1-§6.4 and §6.6.
 - **`resources.json`** — deterministic multi-part Resource ciphertext, part packets, hashmap, advertisement, and proof body. Verified by `tools/verify_resource.py`; regenerated by `tools/regen_resources.py`. Covers SPEC.md §10.2, §10.4, §10.8, and §10.12.
+- **`link-lxmf.json`** — deterministic DIRECT LXMF vectors at the exact PACKET/RESOURCE boundary, using the session key from `links.json`. Verified by `tools/verify_link_lxmf.py`; regenerated by `tools/regen_link_lxmf.py`. Covers SPEC.md §5.2, §5.5, §5.6, §6.4.3, and §10.1.
 
-Remaining vector work should focus on link-delivered LXMF bodies and broader negative/rejection cases rather than the original bootstrap categories.
+Remaining vector work should focus on broader negative/rejection cases rather than the original bootstrap categories.
 
 An implementation that round-trips every test vector — both directions — should be wire-compatible with upstream Reticulum and LXMF for the covered operations.