model-strategy
SolidMulti-model orchestration and model-switching strategy. Score-based model selection, cross-agent delegation (Gemini, Codex, Ollama), advisor pairing, escalation triggers, permission matrix, and cost-efficiency optimization.
Install
Quality Score: 83/100
Skill Content
Details
- Author
- ellmos-ai
- Repository
- ellmos-ai/skills
- Created
- 6 months ago
- Last Updated
- today
- Language
- Python
- License
- MIT
Integrates with
Similar Skills
Semantically similar based on skill content — not just same category
beastmode
Multi-agent orchestration framework for high-intensity feature implementation. Routes work across model tiers: frontier models (Claude Fable, Kimi 3, Opus; Codex only when explicitly selected) own design, architecture, and review sign-off, while economy models (MiniMax M3, Qwen/Gwen) handle implementation and mechanical validation in isolated worktrees, with a self-improving learning loop that promotes lessons back into skills. Harness-agnostic: works with Hermes ACN (async parallel sub-agents), Pi, Claude Code, Codex, Ultraswarm, GSD, delegate_task, or manual orchestration.
claude-model-optimizer
Load before answering whenever a request mentions Claude Opus 5, Claude Fable 5, GPT-5.6 Sol, Terra, or Luna, Codex, or asks which model or agent should plan, implement, review, research, or verify work. Use for model routing, cross-model and dynamic workflows, Codex delegation, effort selection, context preparation, cost controls, and runtime/browser verification. Do not use for ordinary implementation or review with no model-choice question, or for generic prompt rewriting.
multi-model-routing
Use this skill any time a request repeats one operation across many items: summarizing all the logs or documents in a folder, reformatting every file in a directory, tagging or classifying hundreds of rows, reviews, or records, sweeping a codebase for TODOs, or generating dozens-to-hundreds of test records. Phrases like "every file", "all of them", a folder path, or a count ("30 logs", "200 reviews") are the signal -- this is the required way to execute batch work, not an optional optimization, so invoke it before reading any files or writing any code. Also use it whenever another model or agent system enters the picture: a local model engine, OpenCode Zen, Antigravity, Codex/GPT, Gemini, or any non-Claude model -- reaching a model your own harness does not offer, offloading work to save Claude quota, or asking what model backends this machine has. Do not use for single-item analysis, architecture or design, security review, or prose writing.