chore: canonical naming migration

SKILL.md

@@ -1,11 +1,11 @@
 ---
 name: gr00t-n17-libero
 description: >-
-  S1 Vision-Language-Action policy. Capabilities: pick, place, open, close on bowl, cup, drawer, object. NVIDIA Isaac GR00T N1.7 (3B, Cosmos-Reason2-2B VLM backbone) finetuned on the LIBERO benchmark, packaged for OpenRAL. 7-D LIBERO action space (delta end-effector 6-DoF + gripper) over two RGB views. Runs out-of-process via the GR00T ZMQ sidecar (ADR-0046). Open Model License — commercial use permitted. Discovery view of an OpenRAL rSkill — NOT directly runnable by an agent harness; it runs via rSkill.from_pretrained + the robot HAL.
+  S1 Vision-Language-Action policy. Capabilities: pick, place, open, close on bowl, cup, drawer, object. NVIDIA Isaac GR00T N1.7 (3B, Cosmos-Reason2-2B VLM backbone) finetuned on the LIBERO benchmark, packaged for OpenRAL. 7-D LIBERO action space (delta end-effector 6-DoF + gripper) over two RGB views. Runs in-process via lerobot 0.6.0's native GrootPolicy with an NF4-quantized backbone (~5.2 GiB peak, fits an 8 GB GPU). Open Model License — commercial use permitted. Discovery view of an OpenRAL rSkill — NOT directly runnable by an agent harness; it runs via rSkill.from_pretrained + the robot HAL.
 metadata:
   openral_rskill: true            # generated discovery view of an rSkill
   schema_version: 0.1
-  rskill_id: OpenRAL/rskill-gr00t-n17-libero
+  rskill_id: OpenRAL/rskill-gr00t_n17-franka_panda-libero_spatial-bf16
   manifest: ./rskill.yaml
   role: s1
   kind: vla
@@ -24,7 +24,7 @@ metadata:
   latency_budget: {per_chunk_ms: 1500.0}
   license_code: Apache-2.0
   license_weights: nvidia_open_model
-  weights_uri: hf://OpenRAL/rskill-gr00t-n17-libero
+  weights_uri: hf://OpenRAL/rskill-gr00t_n17-franka_panda-libero_spatial-bf16
   source_repo: hf://nvidia/GR00T-N1.7-LIBERO
   paper_url: https://arxiv.org/abs/2503.14734
 ---
@@ -39,7 +39,7 @@ metadata:
 
 ## What it is
 
-An OpenRAL **Vision-Language-Action policy** (`role: s1`, `kind: vla`). NVIDIA Isaac GR00T N1.7 (3B, Cosmos-Reason2-2B VLM backbone) finetuned on the LIBERO benchmark, packaged for OpenRAL. 7-D LIBERO action space (delta end-effector 6-DoF + gripper) over two RGB views. Runs out-of-process via the GR00T ZMQ sidecar (ADR-0046). Open Model License — commercial use permitted.
+An OpenRAL **Vision-Language-Action policy** (`role: s1`, `kind: vla`). NVIDIA Isaac GR00T N1.7 (3B, Cosmos-Reason2-2B VLM backbone) finetuned on the LIBERO benchmark, packaged for OpenRAL. 7-D LIBERO action space (delta end-effector 6-DoF + gripper) over two RGB views. Runs in-process via lerobot 0.6.0's native GrootPolicy with an NF4-quantized backbone (~5.2 GiB peak, fits an 8 GB GPU). Open Model License — commercial use permitted.
 
 ## Capabilities
 
@@ -66,7 +66,7 @@ this file. Execution always goes through the OpenRAL loader and the robot HAL.
 ```python
 from openral_rskill import rSkill
 
-skill = rSkill.from_pretrained("OpenRAL/rskill-gr00t-n17-libero")
+skill = rSkill.from_pretrained("OpenRAL/rskill-gr00t_n17-franka_panda-libero_spatial-bf16")
 # the loader validates embodiment / sensors / runtime / quantization against the target
 # RobotDescription and enforces the weight-license gate before any weights load.
 ```