大师.skill (Master Skill)

What 大师.skill (Master Skill) does

Given an industry name, 大师.skill runs six research tracks, leading practitioners, tool landscape, workflows, canonical knowledge, information sources and terminology, and produces a skill that reasons with that industry's mental models rather than reciting facts. Each generated skill ships with its own cli/ folder containing a research protocol script, decision-tree scripts and workflow walkthroughs, each supporting --help, --explain, --dry-run and --json.

The project has already distilled 72 industries, spanning technical, business, content, medical, legal and craft domains, and borrows five mechanisms from Microsoft's SkillOpt project, including a validation gate and a protected zone that keeps a skill's core judgment from being overwritten by routine updates. Generated skills install into any agent that reads the Agent Skills SKILL.md format.

Key features

• Six-track automated research pipeline per industry (people, tools, workflows, knowledge, sources, terminology)
• Auto-generated cli/ command suite with research-protocol, decision-tree and workflow scripts
• Validation gate and protected-zone editing adapted from Microsoft SkillOpt
• 72 pre-distilled industries ready to install
• Works in any agent that supports the SKILL.md format, including Hermes

When to use it

• Giving an agent the working judgment of a specific industry instead of generic prompts
• Getting a decision-tree CLI for a recurring choice, such as framework selection
• Bootstrapping a new vertical skill pack for a niche profession in under an hour

Who it is for: Builders who want an AI agent to reason like a seasoned practitioner in a specific industry rather than search generic knowledge.

How it fits with Hermes Agent

大师.skill names Hermes Agent as one of the agents a generated skill can be installed into, alongside Claude Code, OpenClaw and Codex.

Note: Primarily documented in Chinese, with an English README available separately.

FAQ