Hi - I answer from the OpenSmartRoute documentation: routing, the API, plans and quotas, self-hosting. Ask away, or open a support ticket if you need a person.
Grounded in the docs - follow a source before acting on it.
gpu-first-execution - Skill - OpenSmartRoute
Skillv1.0.0
gpu-first-execution
Verify and enforce actual NVIDIA CUDA GPU execution for ML training, inference, evaluation, or benchmarking when GPU use is required. Unexpected CPU fallback is a defect.
Imported from shinevijayan03/ProtoLoop-AI (.agents/skills/gpu-first-execution/SKILL.md). Install upstream with npx skills add shinevijayan03/ProtoLoop-AI --skill gpu-first-execution. Copyright stays with the author.
Procedure
Detect NVIDIA GPU and driver.
Verify framework CUDA support.
Select the intended device explicitly.
Run a minimal GPU smoke test.
Run the target workload.
Verify actual model/tensor/device placement.
Inspect VRAM or utilization when useful.
Capture evidence.
cuda.is_available() alone is not proof of workload execution.
Use it
Copy one of these into your project. Installing also returns the manifest and these snippets.
# after Install: the listing is in your workspace's routing pool - a plan picks it for its slot
curl -s -X POST https://api.opensmartroute.ai/api/v1/route -H 'Authorization: Bearer $OSR_API_KEY' -H 'Content-Type: application/json' -d '{"text": "...", "plan": true}'
Manifest
An Open Capability Manifest: the router reads it to know what this does, what it costs and when to pick it.
{
"ocm": "1",
"id": "shinevijayan03-protoloop-ai-gpu-first-execution",
"kind": "skill",
"name": "gpu-first-execution",
"description": "Verify and enforce actual NVIDIA CUDA GPU execution for ML training, inference, evaluation, or benchmarking when GPU use is required. Unexpected CPU fallback is a defect.",
"publisher": "shinevijayan03",
"version": "1.0.0",
"capabilities": {
"domains": [
"general"
],
"tags": [
"skill-md",
"github"
],
"languages": [
"en"
]
},
"quality_prior": 0.6,
"examples": [
"Verify and enforce actual NVIDIA CUDA GPU execution for ML training, inference, evaluation, or benchmarking when GPU use is required. Unexpected CPU fallback is a defect."
],
"primary": false,
"metadata": {
"source": {
"provider": "github",
"repository": "https://github.com/shinevijayan03/ProtoLoop-AI",
"path": ".agents/skills/gpu-first-execution/SKILL.md",
"ref": "34b10a30fcd81f2063b1cc328c4c5d5fd9d2134c",
"url": "https://github.com/shinevijayan03/ProtoLoop-AI/blob/34b10a30fcd81f2063b1cc328c4c5d5fd9d2134c/.agents/skills/gpu-first-execution/SKILL.md",
"key": "shinevijayan03/ProtoLoop-AI/.agents/skills/gpu-first-execution/SKILL.md"
}
},
"instructions": "# Procedure\n1. Detect NVIDIA GPU and driver.\n2. Verify framework CUDA support.\n3. Select the intended device explicitly.\n4. Run a minimal GPU smoke test.\n5. Run the target workload.\n6. Verify actual model/tensor/device placement.\n7. Inspect VRAM or utilization when useful.\n8. Capture evidence.\n\n`cuda.is_available()` alone is not proof of workload execution.",
"cost": {
"context_tokens": 89
}
}
Fetch it by URL: GET /api/v1/registry/shinevijayan03-protoloop-ai-gpu-first-execution/manifest?version=1.0.0
Reviews
Star ratings from people who tried it. One review per account; edit yours any time.
No reviews yet. Install it, try it, and be the first to rate it.
