技能工厂
- 作者仓库星标 191
- 许可证 MIT
- 作者仓库 hermeshub
- 领域
- 通用
- 作者声明的智能体
- 未找到明确声明;不据此推断已兼容或已测试
- 静态检查
- 100 / 100 · 启发式扫描,不代表运行安全
- 运行环境证据
- 未执行;当前没有隔离运行测试证据
- 作者 / 版本 / 许可证
- @amanning3390 · v1.0.0 · MIT
- 上下文用量评级
- 低消耗
- 接入复杂程度
- 即装即用
- 是否需要外部接口密钥
- 不需要
- 兼容的系统
- 未声明(默认跨平台)
- 运行环境要求
- Python >=3.10
- 检测到的文件与系统行为
-
- 只读
- 允许写入 / 修改
- 网络行为
- 仅限本地
- 安装命令数
- 无(仅作为资料)
第一阶段:沉默观察
当处于活动状态时,该技能会维护会话活动日志,而不会将其显示给用户。它跟踪的内容: 重复操作 - 多次使用的任何命令、序列或方法 多步骤工作流程 - 实现连贯目标的 3 个以上步骤的序列 工具组合 - 以一致的模式一起使用两个或多个工具
第 2 阶段:触发条件
当出现以下任一情况时,建议创建技能: 触发器 · 示例用户明确请求 · “将此保存为技能”、“记住此工作流程” 斜线命令 · `/skill-factory propose` 重复模式 (2x+) · 相同的工作流程在会话中出现两次
第三阶段:提案和生成
触发后,Skill Factory 会根据检测到的工作流程提出结构化建议,然后生成完整的技能包: SKILL.md — 以标准 Hermes 技能格式编写的 AI 指令,其中包含阶段、步骤、质量检查表和会话中的真实示例。
# Skill Factory
A meta-skill plugin for Hermes Agent that silently observes user workflows, detects repeatable patterns (e.g., Python env setup, git PR creation), and automatically proposes and generates reusable Hermes skills. Turns lived experience into procedural memory — every workflow you repeat becomes a skill waiting to be born.
## How It Works
Skill Factory operates in three stages:
```
┌──────────────────────────────────────────────────────────┐
│ Your Session │
│ "write a test → run it → fix the failure → commit" │
└────────────────────────┬─────────────────────────────────┘
│ observed silently
▼
┌──────────────────────────────────────────────────────────┐
│ SKILL.md — The Meta-Skill (AI Brain) │
│ Tells Hermes HOW to observe, analyze, and propose skills │
│ Location: ~/.hermes/skills/meta/skill-factory/SKILL.md │
└────────────────────────┬─────────────────────────────────┘
│ proposes + generates
▼
┌──────────────────────────────────────────────────────────┐
│ plugin.py — The Command Interface │
│ /skill-factory propose | list | save | status | clear │
│ Location: ~/.hermes/plugins/skill_factory.py │
└────────────────────────┬─────────────────────────────────┘
│ writes files to disk
▼
┌──────────────────────────────────────────────────────────┐
│ Generated Skill Package │
│ ~/.hermes/skills/<category>/<name>/SKILL.md │
│ ~/.hermes/plugins/<name>.py │
└──────────────────────────────────────────────────────────┘
```
## Phase 1: Silent Observation
While active, the skill maintains a log of session activity without surfacing it to the user.
**What it tracks:**
- Repeated actions — any command, sequence, or approach used more than once
- Multi-step workflows — sequences of 3+ steps that accomplish a coherent goal
- Tool combinations — two or more tools used together in a consistent pattern
- Domain patterns — how the user approaches problems specific to their domain
- Fixes and workarounds — recurring debugging patterns or solutions
**What it ignores:**
- One-off tasks with no reuse potential
- Trivial single-step actions
- Workflows already handled by existing Hermes skills
- Highly context-specific tasks that won't generalize
## Phase 2: Trigger Conditions
Skill creation is proposed when any of the following occur:
| Trigger | Example |
|---|---|
| User explicitly requests | "save this as a skill", "remember this workflow" |
| Slash command | `/skill-factory propose` |
| Repeated pattern (2x+) | Same workflow appeared twice in the session |
| Session winding down | User says "done", "thanks", or wraps up |
| User frustration hint | "I always have to do this manually..." |
## Phase 3: Proposal & Generation
When triggered, Skill Factory presents a structured proposal with the detected workflow, then generates a complete skill package:
- **SKILL.md** — AI instructions written in standard Hermes skill format with phases, steps, quality checklists, and real examples from the session
- **plugin.py** — A slash command that triggers the workflow directly, with hooks and tool registrations scaffolded from the detected pattern
## Commands
| Command | Description |
|---|---|
| `/skill-factory propose` | Analyze current session and propose the top detected skill |
| `/skill-factory list` | Show all skills generated in this session |
| `/skill-factory status` | Show what patterns are currently being tracked |
| `/skill-factory queue` | Show all detected patterns queued for proposal |
| `/skill-factory save <name>` | Name and save the last proposed skill |
| `/skill-factory clear` | Clear the current session tracking log |
## Installation
```bash
git clone https://github.com/Romanescu11/hermes-skill-factory.git
cd hermes-skill-factory
bash install.sh
```
Or manually copy to your Hermes directories:
```bash
cp skills/skill-factory/SKILL.md ~/.hermes/skills/meta/skill-factory/SKILL.md
cp plugins/skill_factory.py ~/.hermes/plugins/skill_factory.py
hermes skills reload
hermes skills enable skill-factory
```
## Requirements
- Hermes Agent v2026.3+
- Python 3.10+
- No external dependencies
## Quality Standards
Generated SKILL.md files must:
- Be actionable with concrete steps, not vague guidance
- Include at least one real example from the triggering session
- Define clear trigger conditions
- Stay under 600 lines
- Capture the reasoning behind each step, not just the mechanics
Generated plugin.py files must:
- Include a docstring with install and usage instructions
- Register at minimum one slash command
- Handle errors gracefully
- Be idiomatic Python with type hints and async/await
章节与文件索引
按说明文件中的顺序排列。
它是如何运作的
第一阶段:沉默观察
第 2 阶段:触发条件
第三阶段:提案和生成
命令
文件/命令 -> /skill-factory propose · /skill-factory list · /skill-factory status · /skill-factory queue · /skill-factory save <name> · /skill-factory clear
内容校验值 -> 8175b10c05c7
当前原文的自动摘译,尚未人工审校。
它是如何运作的
技能工厂分三个阶段运作:
第 2 阶段:触发条件
当发生以下任一情况时,建议创建技能: 触发 · 示例用户明确请求 · “将此保存为技能”、“记住此工作流程” 斜线命令 ·
/skill-factory propose重复模式 (2x+) · 相同的工作流程在会话中出现两次命令
命令 · 描述
/skill-factory propose· 分析当前会话并提出检测到最多的技能/skill-factory list· 显示此会话中生成的所有技能/skill-factory status· 显示当前正在跟踪的模式/skill-factory queue· 显示排队等待建议的所有检测到的模式