fleet-doctrine
由 jarvis-drakon 提供
Web & Frontend Development简介
Model routing strategy for multi-model AI fleet.
描述
该技能的功能
Defines which AI model to use for each type of task in a multi-model agent setup. Maps task categories to specific models: Opus for high-stakes decisions, Codex for coding, Sonnet for routine work, Gemini for vision and long context, Grok for fast secondary opinions.
Having explicit routing rules prevents expensive models from being wasted on trivial tasks and prevents weak models from handling decisions that require judgment.
何时使用
- Choosing a model when spawning a sub-agent for a code refactor
- Setting the default model for scheduled cron jobs
- Deciding whether to escalate a financial decision to Opus
- Running parallel code reviews with Codex and Grok
- Picking the right model for a document with over a million tokens
示例工作流
User asks: which model should handle a nightly email briefing cron job?
- 1
Check if the task is security or finance related — it is not
- 2
Check if it requires complex coding — it does not
- 3
Check if it needs image generation or 1M+ context — it does not
- 4
Identify the task as routine and templated
- 5
Apply the Sonnet routing rule for scheduled/templated work
Assigns Sonnet as the model for the cron job
安装
clawhub install jarvis-drakon/fleet-doctrine运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。