blake-r/dsh-plugins--packages-dsh-skill-from-tools ↗★ 0
@blake-r/dsh-skill-from-tools
为每个工具生成独立技能（不进行模块分组）的简短版 DSH 插件，明确声明工具为直接调用而非通过 MCP 网关路由。
AI 分析
核心用途是防止模型混淆原生工具与 MCP 工具，确保工具通过 direct tool_call 调用。适合在使用多工具混合场景下遇到模型路由混乱的用户。
安装
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:blake-r/dsh-plugins#585c40b34c69d84da20592d9b396bd78b4fafb30&path:packages/dsh-skill-from-tools说明文档
阅读完整 README ↗@blake-r/dsh-skill-from-tools
One skill per tool (no module grouping), short form.
Install
From npm (published):
dsh plugin --profile web add @blake-r/dsh-skill-from-tools
From the GitHub repo (selective, subdirectory):
dsh plugin --profile web add github:blake-r/dsh-plugins#path:packages/dsh-skill-from-tools
What it does
A bundle package in the dsh-plugins monorepo. See the header comment in
src/dsh-skill-from-tools.mjs for full behavior.
Each skill explicitly states that its tool is invoked as a direct tool_call
with its own name — not routed through the mcp gateway — so the model does
not confuse the skill's tool with an MCP tool.