From de1b8f3e8d543deea34c1fa5e006b0c9971038fc Mon Sep 17 00:00:00 2001
From: comfyanonymous <121283862+comfyanonymous@users.noreply.github.com>
Date: Fri, 3 Jul 2026 13:08:24 -0700
Subject: [PATCH] Update AGENTS.md (#14738)

---
 AGENTS.md | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/AGENTS.md b/AGENTS.md
index 5236a00cf..a8bacbd5e 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -171,6 +171,9 @@
 - Reuse existing model classes, blocks, ops, and helper modules when appropriate.
   Before implementing a new version of a model component, search the existing
   model code for a class or helper that already provides the behavior.
+- Model detection code that inspects linear weight shapes should only use the
+  first dimension. The second dimension may be half the original size for
+  NVFP4 or other 4-bit quantized models.
 - Avoid adding `einops` usage in core inference code. Use native torch tensor
   ops such as `reshape`, `view`, `permute`, `transpose`, `flatten`, `unflatten`,
   `unsqueeze`, and `squeeze` instead.