| name | provider-budget-capability |
| description | Use when planning provider/model/tool usage, BYO keys, local runtime availability, dry-run costs, max-cost gates, capability checks, and explicit fallback decisions. |
Provider Budget Capability
Use this skill before generation-heavy workflows.
Report:
- required providers/runtimes;
- installed/local/remote availability;
- BYO key or OAuth requirements;
- estimated cost and time;
- missing models/tools;
- fallback options and their quality tradeoffs.
Rules:
--max-cost 0 should produce a plan but block paid generation.
- Missing local runtime must not silently become cloud runtime.
- Provider/model fallback needs an explicit decision log.
Clash Managed Dry Run
For managed local execution, write a readable request file:
{
"workflowId": "short-drama-episode-001",
"stage": "generate",
"maxCostUsd": 0,
"operations": [
{
"id": "character-three-view",
"capability": "image.generation",
"provider": "local-comfyui",
"runtime": "comfyui",
"mode": "local",
"availability": "available",
"estimatedCostUsd": 0,
"estimatedSeconds": 90,
"requiresByoKey": false
}
],
"fallbackOptions": []
}
Then ask Clash to write the gate artifact:
clash production plan-dry-run-cost-gate \
--request plans/generate.dry-run-request.json \
--out reviews/gates/generate.dry-run-cost-gate.json \
--json
The result uses kind: "clash.workflow.dry-run-cost-gate" and records
availability, estimated cost/time, max-cost blocking, rejected fallback options,
fallbackUsed: false, and executionAllowed. This command must not execute
generation, download, render, or provider calls.