Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
The Smart Spawn API selects the best AI model for a given task based on task type and budget tier. It returns a scored recommendation with pricing and reasoning, then you spawn a sub-agent using the returned model ID. No plugin or account required — plain HTTP requests to ss.deeflect.com/api.
Instead of hardcoding a model in every spawn call, it routes dynamically using benchmark scores refreshed every 6 hours from 5 sources, so recommendations reflect current model capabilities and pricing.
When to use it
- Picking the cheapest viable model for a batch data extraction job
- Routing a vision task to a model with image support before spawning
- Comparing two candidate models side-by-side before committing to one
- Decomposing a multi-step SaaS build into subtasks, each with its own optimal model
- Running parallel competitor research and slide generation as a dependency graph
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Build a React dashboard with authentication
- 1Call GET ss.deeflect.com/api/pick?task=build+a+react+dashboard+with+auth&budget=medium
- 2Receive recommendation: anthropic/claude-opus-4.6 with score 86 and reason string
- 3Call sessions_spawn with task description and model set to the returned ID
- 4Monitor the spawned session for output
A spawned sub-agent running on the recommended model completes the dashboard build