Hermes Agent官方可选装Skills整理
·
Hermes Agent内置了90个Skills, 在官方的Skills Hub中 还有选装的官方70多个Skills, 整理如下（剔除了国内无法使用的），覆盖AI代理、创意生成、DevOps、MLOps、健康、安全、研究等10余类场景，请君自取。
| 名称 (Name) | 分类 (Category) | 简介 (Description) | 地址 (URL) |
| blackbox | 自治AI代理 | 委托编码给 Blackbox AI CLI（多模型+内置评判） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-blackbox |
| honcho | 自治AI代理 | 配置 Honcho 跨会话记忆：用户建模、隔离、会话摘要 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-honcho |
| one-three-one-rule | 沟通 | 结构化决策框架：技术方案和权衡分析 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/communication/communication-one-three-one-rule |
| blender-mcp | 创意 | 通过 socket 连接直接控制 Blender：3D 建模、材质、动画 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/creative/creative-blender-mcp |
| concept-diagrams | 创意 | 生成扁平极简 SVG 概念图（亮/暗模式自适应） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/creative/creative-concept-diagrams |
| hyperframes | 创意 | 创建 HTML 视频合成、动画标题卡、社交叠加、音视频反应 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/creative/creative-hyperframes |
| kanban-video-orchestrator | 创意 | 多 Agent 视频制作流程：叙事片、营销、音乐视频 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/creative/creative-kanban-video-orchestrator |
| meme-generation | 创意 | 生成真实表情包图片（选模板+加文字，输出 .png） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/creative/creative-meme-generation |
| inference-sh-cli | DevOps | 通过 infsh CLI 运行 150+ AI 应用：图像、视频、LLM、搜索 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/devops/devops-cli |
| docker-management | DevOps | 管理 Docker 容器、镜像、卷、网络、Compose 堆栈 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/devops/devops-docker-management |
| adversarial-ux-test | 质量测试 | 扮演最难搞的用户角色测试产品，发现 UX 痛点 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/dogfood/dogfood-adversarial-ux-test |
| agentmail | 邮件 | 给 Agent 自身配专属邮箱：通过 AgentMail 自主收发邮件 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/email/email-agentmail |
| fitness-nutrition | 健康 | 健身计划和营养追踪：690+ 运动、38万+ 食物数据 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/health/health-fitness-nutrition |
| neuroskill-bci | 健康 | 连接 NeuroSkill 获取实时认知/情绪状态（专注力、放松度等） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/health/health-neuroskill-bci |
| fastmcp | MCP | 用 FastMCP 在 Python 中构建/测试/部署 MCP 服务器 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mcp/mcp-fastmcp |
| mcporter | MCP | 用 mcporter CLI 列出、配置、认证、调用 MCP 工具 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mcp/mcp-mcporter |
| openclaw-migration | 迁移 | 将 OpenClaw 自定义配置迁移到 Hermes Agent | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/migration/migration-openclaw-migration |
| huggingface-accelerate | MLOps | 最简分布式训练 API：4 行代码添加分布式支持 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-accelerate |
| chroma | MLOps | 开源嵌入数据库：向量搜索+全文搜索+元数据过滤 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-chroma |
| clip | MLOps | OpenAI CLIP：零样本图像分类、图文匹配、跨模态检索 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-clip |
| faiss | MLOps | Facebook 高效相似性搜索库：支持十亿级向量、GPU 加速 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-faiss |
| optimizing-attention-flash | MLOps | Flash Attention：2-4x 加速、10-20x 显存节省 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-flash-attention |
| guidance | MLOps | 用正则/语法控制 LLM 输出：保证有效 JSON/XML/代码 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-guidance |
| hermes-atropos-environments | MLOps | 构建 Hermes Agent RL 环境用于 Atropos 训练 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-hermes-atropos-environments |
| huggingface-tokenizers | MLOps | 快速 Tokenizer：Rust 实现、20秒处理 1GB（BPE/WordPiece/Unigram） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-huggingface-tokenizers |
| instructor | MLOps | 从 LLM 响应中提取结构化数据：Pydantic 验证、自动重试 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-instructor |
| lambda-labs-gpu-cloud | MLOps | 按需/预留 GPU 云实例：SSH 访问、持久存储、多节点集群 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-lambda-labs |
| llava | MLOps | LLaVA：视觉指令微调、图像对话、视觉问答 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-llava |
| modal-serverless-gpu | MLOps | 无服务器 GPU 云平台：按需 GPU、部署 ML 模型 API | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-modal |
| nemo-curator | MLOps | GPU 加速数据整理：模糊去重(16x)、质量过滤、PII 检测 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-nemo-curator |
| peft-fine-tuning | MLOps | PEFT 参数高效微调：LoRA、QLoRA、25+ 方法 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-peft |
| pinecone | MLOps | 托管向量数据库：混合搜索、元数据过滤、<100ms 延迟 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-pinecone |
| pytorch-fsdp | MLOps | PyTorch FSDP 全分片数据并行训练：参数分片、混合精度、CPU 卸载 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp |
| pytorch-lightning | MLOps | 高级 PyTorch 框架：Trainer、自动分布式训练、回调系统 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-pytorch-lightning |
| qdrant-vector-search | MLOps | 高性能向量相似性搜索：Rust 引擎、混合搜索、生产级 RAG | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-qdrant |
| sparse-autoencoder-training | MLOps | 用 SAELens 训练/分析稀疏自编码器：解释神经网络特征 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-saelens |
| simpo-training | MLOps | SimPO 简单偏好优化：无参考模型的 DPO 替代方案 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-simpo |
| slime-rl-training | MLOps | LLM 后训练 RL：Megatron+SGLang 框架 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-slime |
| stable-diffusion-image-generation | MLOps | Stable Diffusion 文生图：Diffusers、图生图、修复 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-stable-diffusion |
| tensorrt-llm | MLOps | NVIDIA TensorRT 优化 LLM 推理：10-100x PyTorch 加速 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-tensorrt-llm |
| distributed-llm-pretraining-torchtitan | MLOps | PyTorch 原生分布式 LLM 预训练：4D 并行（FSDP2/TP/PP/CP） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-torchtitan |
| whisper | MLOps | OpenAI Whisper 语音识别：99 种语言、转录、翻译 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/mlops/mlops-whisper |
| canvas | 效率 | Canvas LMS 集成：获取课程和作业 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-canvas |
| here.now | 效率 | 发布静态网站到 here.now + 云端文件存储 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-here-now |
| memento-flashcards | 效率 | 间隔重复闪卡系统：从笔记/YouTube 生成测验 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-memento-flashcards |
| shop-app | 效率 | Shop.app：产品搜索、订单追踪、退货 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-shop-app |
| shopify | 效率 | Shopify GraphQL API：产品、订单、客户、库存、元字段 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-shopify |
| siyuan | 效率 | 思源笔记 API：搜索/读取/创建笔记块和文档（中国产品） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-siyuan |
| telephony | 效率 | 给 Hermes 添加电话能力：Twilio SMS/MMS/通话、Bland.ai 外呼 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/productivity/productivity-telephony |
| bioinformatics | 研究 | 网关到 400+ 生物信息学技能：基因组学、转录组学、单细胞 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-bioinformatics |
| domain-intel | 研究 | 被动域名侦察：子域名发现、SSL 证书、WHOIS、DNS | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-domain-intel |
| drug-discovery | 研究 | 药物发现助手：ChEMBL 搜索、ADMET 分析、药物相似性 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-drug-discovery |
| duckduckgo-search | 研究 | 免费 DuckDuckGo 搜索：文本/新闻/图片/视频（无需 API 密钥） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-duckduckgo-search |
| gitnexus-explorer | 研究 | 用 GitNexus 索引代码库：交互式知识图谱 + Web UI | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-gitnexus-explorer |
| parallel-cli | 研究 | Parallel CLI：Agent 原生网络搜索、提取、深度研究 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-parallel-cli |
| qmd | 研究 | 本地搜索个人知识库：BM25 + 向量搜索 + LLM 重排序 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-qmd |
| scrapling | 研究 | 网页抓取：HTTP 获取、隐身浏览器、Cloudflare 绕过、爬虫 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-scrapling |
| searxng-search | 研究 | 免费 SearXNG 元搜索：聚合 70+ 搜索引擎（可自建/无需 API 密钥） | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/research/research-searxng-search |
| 1password | 安全 | 设置和使用 1Password CLI：安装、认证、读取/注入密钥 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/security/security-1password |
| oss-forensics | 安全 | 供应链调查：GitHub 仓库取证分析、删除提交恢复、IOC 提取 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/security/security-oss-forensics |
| sherlock | 安全 | OSINT 用户名搜索：400+ 社交网络 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/security/security-sherlock |
| page-agent | Web开发 | 嵌入阿里巴巴 page-agent：纯 JS 页面内 GUI Agent 组件 | https://hermes-agent.nousresearch.com/docs/user-guide/skills/optional/web-development/web-development-page-agent |
AtomGit 是由开放原子开源基金会联合 CSDN 等生态伙伴共同推出的新一代开源与人工智能协作平台。平台坚持“开放、中立、公益”的理念，把代码托管、模型共享、数据集托管、智能体开发体验和算力服务整合在一起，为开发者提供从开发、训练到部署的一站式体验。
更多推荐
5
0- 0
已为社区贡献2条内容
所有评论(0)