diff --git a/README.md b/README.md
index aa76482..5b75a46 100644
--- a/README.md
+++ b/README.md
@@ -13,21 +13,31 @@ Included: checked GGUF loading, safetensors conversion, DDIM sampling, C/C++
 APIs, CPU/Vulkan parity tests, and a local text-to-motion demo. Constraints,
 SOMA, G1, GLB export, and quantised models are not implemented yet.
 
-## Build and test
+## Build and test on Linux
 
-GGML is a pinned Git submodule:
+Install a C++23 compiler, CMake 3.25+, Ninja, Python 3 with the Hugging Face
+CLI (`pip install huggingface_hub`), and the Vulkan loader/headers for Vulkan
+support. GGML is a pinned Git submodule:
 
 ```sh
 git submodule update --init --recursive
-nix develop path:. --command cmake --preset debug
-nix develop path:. --command cmake --build --preset debug
-nix develop path:. --command ctest --preset debug
+scripts/download_gguf_weights.sh --output "$PWD"
+cmake --preset debug
+cmake --build --preset debug
+ctest --preset debug
 ```
 
 The standard test suite requires the local motion GGUF, text bundle, and
 fixtures. It never downloads weights by itself. `release`, `asan-ubsan`, and
 `fuzz` presets are also available.
 
+Nix is optional and provides these dependencies reproducibly:
+
+```sh
+nix develop path:. --command cmake --preset debug
+nix develop path:. --command cmake --build --preset debug
+```
+
 For sanitizer work:
 
 ```sh
@@ -71,7 +81,7 @@ text encoder and the upstream-linked
 diffusion model are separate, so users download rather than recreate them:
 
 ```sh
-nix develop path:. --command scripts/download_gguf_weights.sh --output "$PWD"
+scripts/download_gguf_weights.sh --output "$PWD"
 ```
 
 The installer verifies each published manifest and SHA-256 hashes. Use
diff --git a/scripts/hf/Kimodo-SMPLX-RP-v1-GGML/README.md b/scripts/hf/Kimodo-SMPLX-RP-v1-GGML/README.md
index e7883cd..ed0f37a 100644
--- a/scripts/hf/Kimodo-SMPLX-RP-v1-GGML/README.md
+++ b/scripts/hf/Kimodo-SMPLX-RP-v1-GGML/README.md
@@ -15,10 +15,10 @@ This repository contains only the diffusion model; its reusable Llama-derived
 text encoder is distributed separately as
 [`Llama-3-Kimodo-GGML`](https://huggingface.co/LocalAI-io/Llama-3-Kimodo-GGML).
 
-From a kimodo.cpp checkout, install both with:
+From a kimodo.cpp checkout with the Hugging Face CLI installed, install both with:
 
 ```sh
-nix develop path:. --command scripts/download_gguf_weights.sh --output "$PWD"
+scripts/download_gguf_weights.sh --output "$PWD"
 ```
 
 The model is installed at `models/kimodo-smplx-rp-v1-f32.gguf`. Use
diff --git a/scripts/hf/Llama-3-Kimodo-GGML/README.md b/scripts/hf/Llama-3-Kimodo-GGML/README.md
index 3be0911..657c43f 100644
--- a/scripts/hf/Llama-3-Kimodo-GGML/README.md
+++ b/scripts/hf/Llama-3-Kimodo-GGML/README.md
@@ -11,10 +11,10 @@ reusable LLM2Vec encoder only; download a matching Kimodo diffusion model
 separately, for example
 [`Kimodo-SMPLX-RP-v1-GGML`](https://huggingface.co/LocalAI-io/Kimodo-SMPLX-RP-v1-GGML).
 
-From a kimodo.cpp checkout, install both with:
+From a kimodo.cpp checkout with the Hugging Face CLI installed, install both with:
 
 ```sh
-nix develop path:. --command scripts/download_gguf_weights.sh --output "$PWD"
+scripts/download_gguf_weights.sh --output "$PWD"
 ```
 
 The components intentionally remain split into individual layers so kimodo.cpp