call-models

仓库创建 2026年7月27日最近提交 4 小时前SkillHot 收录 3 小时前

这个 Skill 做什么

▸ 展开 SKILL.md 英文原文

Resolve or call registered AI models from Codex, Claude Code, Kimi Code, Gemini CLI, Antigravity, or another compatible agent, including Sol, Luna, Terra, Opus, Fable, Kimi, and Gemini. Enforce explicit effort, read/write mode, timeouts, recursion and fan-out limits, structured output, subscription-auth checks, and MCP or CLI dispatch. Use when the user names a target model or effort, asks one model to call another, wants cross-vendor work or review, asks which model should handle a task, or builds a multi-model pipeline.

Stars

Forks

仓库内 Skill

积累中

7 日增星

安装 / 使用

给你的 Agent 一句话（通用）

帮我安装这个 skill：https://raw.githubusercontent.com/21mlimit/call-models/main/skills/call-models/SKILL.md

或 curl 直取 SKILL.md

curl -fsSL "https://raw.githubusercontent.com/21mlimit/call-models/main/skills/call-models/SKILL.md"

SKILL.MD 节选查看完整文件 ↗

# Call models

Resolve and dispatch model calls deterministically. Preserve an explicit user
model and effort. Default to read-only.

## Resolve without loading the registry

Do not read `models.json` into model context. The router validates it
deterministically.

Prefer the configured `call-models` MCP tools:

1. Call `resolve_model` when deciding between native and CLI dispatch.
2. Call `call_model` for cross-runtime work or whenever exact effort must be
   enforced.
3. Call `routing_policy` when the user did not choose a model or the task can
   affect order execution, P&L, reconciliation, broker auth, or financial
   correctness.
4. Call `doctor` when auth, availability, or effort suppor

via SKILL·HOT · 数据来自 GitHub 公开信息 · 原文版权归作者所有