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.
finops-ai-ml-costs - Skill - OpenSmartRoute
Skillv1.0.0
finops-ai-ml-costs
See `content/06-decision-tree.xml`. The tree starts from a concrete observable signal and routes each branch to a `<conclusion ref="rule-id">` resolved against `content/01-core-rules.xml`. Use it when
Imported from faionfaion/faion-network (skills/faion/knowledge/infra/finops-ai-ml-costs/AGENTS.md). Install upstream with npx skills add faionfaion/faion-network --skill finops-ai-ml-costs. Copyright stays with the author.
FinOps for AI / ML Workloads
Summary
One-sentence: Generates an AI/ML cost-control config (GPU spot strategy with checkpointing + inference caching + quantization + batch policy + per-run/per-request attribution) for training and inference.
One-paragraph: Generates an AI/ML cost-control config (GPU spot strategy with checkpointing + inference caching + quantization + batch policy + per-run/per-request attribution) for training and inference. The methodology pins the artefact shape, ties every conclusion to a rule, and routes the operator via a decision tree that always terminates either on an applicable rule or on skip-this-methodology. Apply when preconditions hold; skip via the tree otherwise.
Ефективно для:
GPU training pipelines з runtime ≥1h і checkpointable jobs.
Inference fleet з cacheable prompts (RAG, classification, embeddings).
Multi-tenant ML platforms де per-team chargeback потрібен.
Pre-deployment review нового model: cost-per-prediction must clear ROI gate.
Applies If (ALL must hold)
GPU spend ≥$5k/month OR cloud LLM inference spend ≥$2k/month.
Training jobs support checkpoint/resume.
Per-team or per-product attribution required by finance.
Inference traffic has measurable repeat-input rate (≥10%) to justify caching.
Skip If (ANY kills it)
GPU spend <$2k/month — vendor pricing optimization gives bigger ROI than spot.
Training jobs cannot checkpoint (research scripts in flux) — spot is unsafe.
Validate produced artefact against the schema in content/02-output-contract.xml
CI on each artefact change; pre-commit; --self-test in unit run
Related
Parent: pro/infra/cicd-engineer/AGENTS.md
[[finops-framework]]
[[gitops-core-principles]]
Decision tree
See content/06-decision-tree.xml. The tree starts from a concrete observable signal and routes each branch to a <conclusion ref="rule-id"> resolved against content/01-core-rules.xml. Use it whenever you are unsure whether this methodology applies — the tree always terminates either on an applicable rule or on skip-this-methodology.
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": "faionfaion-faion-network-finops-ai-ml-costs",
"kind": "skill",
"name": "finops-ai-ml-costs",
"description": "See `content/06-decision-tree.xml`. The tree starts from a concrete observable signal and routes each branch to a `<conclusion ref=\"rule-id\">` resolved against `content/01-core-rules.xml`. Use it whenever you are unsure whether this methodology applies — the tree always terminates either on an applicable rule or on `skip-this-methodology`.",
"publisher": "faionfaion",
"version": "1.0.0",
"capabilities": {
"domains": [
"general"
],
"tags": [
"skill-md",
"github-agents"
],
"languages": [
"en"
]
},
"quality_prior": 0.6,
"examples": [
"See `content/06-decision-tree.xml`. The tree starts from a concrete observable signal and routes each branch to a `<conclusion ref=\"rule-id\">` resolved against `content/01-core-rules.xml`. Use it whenever you are unsure whether this methodology applies — the tree always terminates either on an applicable rule or on `skip-this-methodology`."
],
"primary": false,
"metadata": {
"source": {
"provider": "github-agents",
"repository": "https://github.com/faionfaion/faion-network",
"path": "skills/faion/knowledge/infra/finops-ai-ml-costs/AGENTS.md",
"ref": "8e31684537e64d45c75264b1e0a2384f9b024a34",
"url": "https://github.com/faionfaion/faion-network/blob/8e31684537e64d45c75264b1e0a2384f9b024a34/skills/faion/knowledge/infra/finops-ai-ml-costs/AGENTS.md",
"key": "faionfaion/faion-network/skills/faion/knowledge/infra/finops-ai-ml-costs/AGENTS.md"
}
},
"instructions": "# FinOps for AI / ML Workloads\n\n## Summary\n\n**One-sentence:** Generates an AI/ML cost-control config (GPU spot strategy with checkpointing + inference caching + quantization + batch policy + per-run/per-request attribution) for training and inference.\n\n**One-paragraph:** Generates an AI/ML cost-control config (GPU spot strategy with checkpointing + inference caching + quantization + batch policy + per-run/per-request attribution) for training and inference. The methodology pins the artefact shape, ties every conclusion to a rule, and routes the operator via a decision tree that always terminat",
"cost": {
"context_tokens": 1030
}
}
Fetch it by URL: GET /api/v1/registry/faionfaion-faion-network-finops-ai-ml-costs/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.
