2026 Hermes Agent Skills 进阶全攻略：SKILL.md、Bundle、GEPA 自进化与 Tap 发布实战

---
name: my-skill                    # 必填：小写字母+连字符，≤64字符
description: |                    # 必填：≤1024字符，建议以"Use when..."开头
  Use when the user needs to [...].
  Handles [...] and [...].
version: 1.0.0                    # 推荐填写
license: MIT
compatibility: Requires git, docker  # 环境依赖说明
allowed-tools: Bash(git:*) Read   # 预授权工具（实验性）
metadata:
  hermes:
    tags: [devops, automation]
    category: software-development
    related_skills: [github-pr-workflow, test-driven-development]
    requires_toolsets: [terminal]
    fallback_for_toolsets: [web]
---

# My Skill Title

## Overview
1-2 paragraphs: what it does and why it exists.

## When to Use
- Use for: [具体场景]
- Don't use for: [明确的排除场景]

## Procedure
1. 步骤一（包含精确命令）
2. 步骤二
3. 步骤三

## Common Pitfalls
1. 常见问题及修复方案
2. 边界情况处理

## Verification Checklist
- [ ] 验证点 1
- [ ] 验证点 2

~/.hermes/skills/
└── my-category/
    └── my-skill/
        ├── SKILL.md              # 主文件（核心步骤，建议 ≤500 行）
        ├── references/
        │   ├── api-docs.md       # API 参考（按需加载）
        │   └── examples.md       # 示例输入输出
        ├── templates/
        │   └── config.yaml       # 可复用模板
        └── scripts/
            └── setup.sh          # Agent 可直接执行的脚本

# ~/.hermes/skill-bundles/backend-dev.yaml
name: backend-dev
description: |
  Full backend feature workflow — code review, TDD, and PR management.
  Use at the start of any new backend feature session.
skills:
  - github-code-review
  - test-driven-development
  - github-pr-workflow
instruction: |
  Always write failing tests first before implementation.
  Open PRs with co-author tags for pair-programming sessions.
  Never push directly to main.

name: research-session
description: Load all research tools at once for deep-dive sessions.
skills:
  - arxiv
  - deep-research
  - plan
  - excalidraw
instruction: |
  Start every session by checking recent papers on the topic.
  Create an Excalidraw diagram for any architecture discussed.

name: mlops-deploy
description: Model deployment pipeline with monitoring setup.
skills:
  - vllm
  - llama-cpp
  - github-pr-workflow
  - systematic-debugging
instruction: |
  Always run inference benchmarks before and after deployment.
  Document model quantization settings in PR description.

hermes bundles create backend-dev \
  --skills github-code-review,test-driven-development,github-pr-workflow \
  --instruction "Always write failing tests first"

hermes bundles list
hermes bundles show backend-dev
/bundles                    # 在对话内查看

metadata:
  hermes:
    requires_toolsets: [web]          # 仅当 web 工具集可用时显示
    requires_tools: [web_search]      # 仅当 web_search 工具可用时显示
    fallback_for_toolsets: [browser]  # 当 browser 工具集不可用时才显示
    fallback_for_tools: [browser_navigate]  # 当 browser_navigate 不可用时才显示

# duckduckgo-search/SKILL.md
name: duckduckgo-search
description: Search the web using DuckDuckGo. Use when web_search is unavailable.
metadata:
  hermes:
    fallback_for_tools: [web_search]   # 有付费 web_search 时自动隐藏

# telegram-notify/SKILL.md
metadata:
  hermes:
    requires_toolsets: [messaging]    # 仅在有消息平台工具时显示
    platforms: [telegram, discord]    # 平台白名单

# 安装官方可选技能
hermes skills install official/research/arxiv

# 从 HTTP URL 直接安装单文件技能
hermes skills install https://example.com/SKILL.md --name my-skill

# 从 GitHub 仓库安装
hermes skills install github:openai/skills/k8s

# 添加自定义 Tap（订阅整个技能仓库）
hermes skills tap add github:my-org/my-skills

hermes skills browse
hermes skills search kubernetes
hermes skills check
hermes skills update

my-skills-tap/
├── skills.sh.json              # 分类配置（可选，影响 Skills Hub 展示）
├── mlops/
│   ├── vllm-deploy/
│   │   └── SKILL.md
│   └── model-benchmark/
│       └── SKILL.md
├── research/
│   ├── paper-summarizer/
│   │   └── SKILL.md
│   └── citation-finder/
│       └── SKILL.md
└── README.md

{
  "groupings": [
    {
      "title": "MLOps & Model Deployment",
      "skills": ["vllm-deploy", "model-benchmark"]
    },
    {
      "title": "AI Research Workflows",
      "skills": ["paper-summarizer", "citation-finder"]
    }
  ]
}

# 团队成员一键订阅
hermes skills tap add github:your-org/your-skills-tap

# 私有仓库（需要 GitHub Token）
hermes skills tap add github:your-org/private-skills --token $GH_TOKEN

# 更新所有 Tap 技能
hermes skills tap update

# 查看已订阅的 Tap
hermes skills tap list

# 在 ~/.hermes/skills/ 初始化 git 仓库
cd ~/.hermes/skills && git init
git remote add origin https://github.com/your-org/personal-skills
git push -u origin main

# 跨设备同步
git pull && hermes skills reset  # 同步后重建内置技能

git clone https://github.com/NousResearch/hermes-agent-self-evolution
cd hermes-agent-self-evolution
pip install -r requirements.txt

export HERMES_AGENT_PATH=~/.hermes

# 合成数据进化（入门推荐）
python -m evolution.skills.evolve_skill \
    --skill github-code-review \
    --iterations 10 \
    --eval-source synthetic

# 真实会话数据（效果更好）
python -m evolution.skills.evolve_skill \
    --skill github-code-review \
    --iterations 10 \
    --eval-source sessiondb

# 联合多来源轨迹（实验性）
python -m evolution.skills.evolve_skill \
    --skill github-code-review \
    --iterations 10 \
    --eval-source mixed \
    --trace-dirs ~/.claude/traces,~/.hermes/sessions

# 加载插件技能（命名空间格式）
skill_view("superpowers:writing-plans")

# 加载时会自动展示同插件下的兄弟技能
# Agent: "This plugin also includes: superpowers:editing, superpowers:research"

name: my-hermes-plugin
skills:
  - name: writing-plans
    path: skills/writing-plans/SKILL.md
  - name: editing
    path: skills/editing/SKILL.md

# ❌ 太模糊，LLM 可能在不相关场景下加载
description: Helps with code.

# ✅ 明确触发条件，精确激活
description: |
  Use when reviewing a pull request, checking for code quality issues,
  security vulnerabilities, or style violations. Handles GitHub PR URLs
  and local git diff output. Do NOT use for writing new code.

## Common Pitfalls

1. **CSS selector brittleness**: Dynamic class names like `css-abc123` change
   on rebuild. Fix: Use data-testid attributes or ARIA roles instead.

2. **Rate limiting on GitHub API**: Burst of >5000 requests/hour returns 403.
   Fix: Add `--rate-limit 100` flag; check `X-RateLimit-Remaining` header.

3. **Token overflow on large diffs**: PRs with >500 files may hit context limits.
   Fix: Use `skill_view("github-code-review", "references/chunking.md")`
   to load the chunking strategy.

## Procedure

3. Run the extraction script:

   The agent will execute: `scripts/extract_schema.py --input $FILE`

   If the script fails, load the manual fallback:
   `skill_view("db-migration", "references/manual-extract.md")`

# Agent 可通过 skill_manage 工具动态维护技能
skill_manage(
    action='patch',
    name='github-code-review',
    old_string='Check for obvious bugs',
    new_string='Check for: null pointer dereferences, SQL injection, XSS vulnerabilities, and obvious logic errors'
)

# 创建全新技能
skill_manage(
    action='create',
    name='my-new-skill',
    content="""---
name: my-new-skill
description: Use when [...].
---
# My New Skill
...""",
    category='automation'
)

skills:
  agent_writes_require_approval: true  # Agent 写入技能前必须人工确认

name: blog-workflow
description: Full tech blog writing workflow.
skills:
  - seo-keyword-research
  - outline-generator
  - code-example-validator
  - bilingual-checker
  - publish-to-platform
instruction: |
  Always research SEO keywords before writing.
  Ensure all code examples are tested and runnable.
  Generate both Chinese and English title options.

---
name: seo-keyword-research
description: |
  Use when planning a technical blog post. Researches search volume,
  competition, and related queries for both Chinese (Baidu, 微信) and
  English (Google, Dev.to) audiences.
metadata:
  hermes:
    requires_toolsets: [web]
    tags: [seo, blogging, content]
---

## When to Use
Use at the start of any blog writing session, before creating the outline.

## Procedure
1. Identify the primary topic (provided by user or inferred from context)
2. Search for Chinese long-tail variations: use "X 怎么用", "X 教程", "X 最佳实践"
3. Search for English long-tail variations: "X tutorial", "how to X", "X vs Y"
4. Cross-reference with platform trends (掘金热榜, Dev.to trending, HN)
5. Output a keyword matrix with 3-5 primary + 10-15 long-tail per language

## Common Pitfalls
- Chinese and English audiences search for the same concept differently
- Technical terms may have different Chinese translations across platforms
  (e.g., "Agent" vs "智能体" vs "代理" — check which is dominant on target platform)