Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This skill provides a framework for managing AI agent costs by matching task complexity to model capability. It distinguishes between routine janitorial work—such as file reads and status checks—and problems requiring deep reasoning, ensuring premium models are reserved for high-complexity tasks.
Developers can use this guidance to determine when to spawn sub-agents or switch models based on specific requirements, including distinctions between text-only and vision-capable models.
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Installs: Adds the documented model-routing guidance to CLAUDE.md or project instructions; it does not install the complete SKILL.md file. · Claude Code, Codex
Before you start
Access to the agent's CLAUDE.md or project instructions.
Manually add this documented block to CLAUDE.md or the applicable project instructions:
## Model Routing
Before executing tasks, classify complexity:
- ROUTINE (file ops, lookups, formatting) → Use cheapest model
- MODERATE (code, summaries, analysis) → Use mid-tier model
- COMPLEX (debugging, architecture, failures) → Use premium model
When spawning sub-agents, default to cheap models unless task requires more.
Installs: Provides a manual reference to the complete model-hierarchy skill's classification rules and integration examples; no specific installation procedure is documented in the supplied evidence. · Other Agent Systems
Review SKILL.md for the full classification rules and integration examples, then follow only the instructions applicable to the target agent system.
A small list of task descriptions replacing <paste a small list of tasks here>
Whether any task requires image or other visual input, if that is not clear from its description
Use the model-hierarchy skill to classify each of these task descriptions as routine, moderate, or complex and recommend the cheapest suitable model tier for each. Briefly explain each classification. Flag any task involving screenshots, photos, charts, or other image input so it is not assigned to a text-only model. Task descriptions: <paste a small list of tasks here>.
Claude
Codex
Gemini CLI
OpenClaw
DeepSeek