经 AI Skill Hub 精选评估，SharpClaw 获评「强烈推荐」。这款Agent工作流在功能完整性、社区活跃度和易用性方面表现出色，AI 评分 8.0 分，适合有一定技术背景的用户使用。
📚 深度解析
SharpClaw 工作流的设计遵循"最小配置，最大复用"原则：核心逻辑已经封装好，用户只需配置自己的 API Key 和业务参数即可快速上手。工作流内置错误处理和重试机制，在网络波动或 API 限速等情况下仍能稳定运行，适合作为生产环境的自动化基础设施。
在实际部署时，建议先在测试环境中运行 3-5 次，验证各个环节的输出结果符合预期，再部署到生产环境。AI Skill Hub 评分 8.0 分，是同类 Agent 工作流中的精选推荐。
📋 工具概览
SharpClaw 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排，将复杂的多步骤任务拆解为清晰的自动化流程，实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成，适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
📖 中文文档
SharpClaw 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排，将复杂的多步骤任务拆解为清晰的自动化流程，实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成，适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
- 可视化 Agent 工作流编排，无需编写复杂代码
- 支持多步骤自动化任务链，实现全流程无人值守
- 与外部 API、数据库和第三方服务无缝集成
- 内置错误处理与自动重试机制，保障稳定运行
- 提供可复用的自动化模板，快速在同类场景部署
- 自动化日常重复性工作，将精力集中于创造性任务
- 构建数据采集 → 处理 → 输出的完整自动化管线
- 实现跨平台、跨系统的数据流转和业务协同
# 克隆仓库 git clone https://github.com/mkn8rn/SharpClaw cd SharpClaw # 查看安装说明 cat README.md # 按 README 完成环境依赖安装后即可使用
- 访问 GitHub 仓库获取工作流文件
- 在对应平台（Dify / Flowise / Make 等）中找到「导入工作流」功能
- 上传工作流文件
- 按照提示配置必要的环境变量和 API Key
- 运行测试确认流程正常后投入使用
# 查看帮助 sharpclaw --help # 基本运行 sharpclaw [options] <input> # 详细使用说明请查阅文档 # https://github.com/mkn8rn/SharpClaw
# sharpclaw 配置说明 # 查看配置选项 sharpclaw --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量（覆盖配置文件） export SHARPCLAW_CONFIG="/path/to/config.yml"
SharpClaw
Alpha. Single developer, early stage, lots of untested surface area. Read the full disclaimer.
Developer notice. The developer of SharpClaw has been hired to work on an undisclosed commercial enterprise tool inspired by this project. Future development on SharpClaw will be limited by time and contractual constraints, but the project is not abandoned. SharpClaw remains open-source under AGPL-3.0, and that will not change regardless of what happens with the commercial version.
SharpClaw is a local AI agent runtime for building agents that can use real capabilities without collapsing everything into one ungoverned shell. It combines provider abstraction, hot-loadable modules, resumable tasks, permissioned tool execution, persistence, audit trails, and optional desktop or gateway frontends in one .NET platform.
It can be a permissive solo coding agent, a headless automation server, an enterprise control plane, a multi-agent workspace, or the runtime behind a customer-facing AI application. The point is not to make every action painful. The point is that broad trust and narrow trust can coexist.
| Layer | What It Does | Why It Matters |
|---|---|---|
| Providers | Normalizes OpenAI-compatible APIs, Anthropic, Google, Ollama, LlamaSharp, and custom providers. | Applications keep the same agents, tasks, tools, and permissions while models change. |
| Modules | Hot-loads providers, tools, resources, task steps, gateway routes, metrics, and frontend hooks. | SharpClaw can gain new capabilities without rewriting Core. |
| Tasks | Runs resumable workflows with agent calls, tool calls, loops, branches, shared data, logs, and output streams. | Many intelligent workflows can live inside SharpClaw instead of needing a separate backend. |
| Permissions | Combines roles, resource grants, tool awareness, channel policy, and optional approval. | Trusted agents can be broad; focused agents and user-facing agents can be tightly contained. |
| Persistence | Supports encrypted JSON-file persistence and EF Core-backed relational persistence projects. | Local experimentation and higher-performance enterprise deployments can use the same contracts. |
| Frontends | Exposes Core through the API, optional Gateway, and modular Uno desktop shell. | SharpClaw can run headless, behind a public proxy, or as an extensible local app. |
Add Your Own Features
SharpClaw is designed so new capabilities do not have to wait for Core to grow them. If you need Computer Use, you can add a module that owns the screen, mouse, keyboard, capture, policy, and approval surface you actually trust. If you need integration with a specific program, that module can wrap its API, CLI, IPC channel, COM surface, file format, or desktop automation path and expose it as typed tools and task steps. Once it exists, the same capability can be shared with other users instead of being trapped in one local script.
That is what makes modules more than mods. A module can contribute providers, tools, resources, task steps, triggers, gateway endpoints, metrics, and desktop UI hooks. It can be a small adapter around a single command, or it can be a whole product surface with its own routes, screens, background workers, and agent-facing contract. In practice, a SharpClaw module can be an integration, a policy layer, an automation pack, a development tool, or a complete web app that happens to run inside the same permissioned agent runtime.
The result is a runtime that can stay small at the center while the edge becomes highly specific. A hospital, game studio, law office, robotics lab, or solo developer can each add the tools and workflows that match their world without turning every agent into an unrestricted shell. SharpClaw gives those features a place to live, a permission model to stand behind, and a way to be reused by people who need the same capability later.
Getting Started
dotnet build SharpClaw.Application.API/SharpClaw.Application.API.csproj
dotnet run --project SharpClaw.Application.API/SharpClaw.Application.API.csprojThen launch the Uno client, create the first admin account, add a provider, sync models, create an agent, open a channel, and start chatting. Local environment templates live under each process's Environment folder and are copied into place when no local .env exists.
Tasks And Modules
Tasks are SharpClaw's application layer. A task can watch an event, call an agent, invoke tools, branch on output, store shared data, pause, resume, stream progress, and return structured results. A compliance review, release assistant, support triage flow, document intake worker, or build-and-test loop can be a task instead of a separate service.
Modules are how those tasks and agents get new abilities. A module can expose Computer Use, a browser, a terminal with custom rules, a medical device API, a Visual Studio or VS Code bridge, a Neovim adapter, a legacy desktop app, or a provider that is not OpenAI-compatible. A sufficiently trusted agent can even help build and hot-load new modules while SharpClaw is running.
高质量的跨平台AI工作流管理应用
- 构建多智能体协作系统的 Agent 开发者
- 本地部署优先选 GGUF 量化模型，节省显存并保持响应速度
- Agent 任务先做 dry-run 验证工具调用链，再开启自主执行
- API key 直接提交到 git 仓库（请用 .env 并加入 .gitignore）
- 显存不足直接 OOM — 优先降低 context 或换更小的量化模型
- CLI：直接 npm install -g / pip install，命令行调用
- 本地部署：CPU 8GB 起，GPU 推荐 16GB+ 显存
- 云端托管：可放在 Vercel / Railway / Fly.io 等 PaaS 平台
⚡ 核心功能
- 可视化 Agent 工作流编排，无需编写复杂代码
- 支持多步骤自动化任务链，实现全流程无人值守
- 与外部 API、数据库和第三方服务无缝集成
- 内置错误处理与自动重试机制，保障稳定运行
- 提供可复用的自动化模板，快速在同类场景部署
- 构建多智能体协作系统的 Agent 开发者
- 本地部署优先选 GGUF 量化模型，节省显存并保持响应速度
- Agent 任务先做 dry-run 验证工具调用链，再开启自主执行
- API key 直接提交到 git 仓库（请用 .env 并加入 .gitignore）
- 显存不足直接 OOM — 优先降低 context 或换更小的量化模型
👥 适合人群
🎯 使用场景
- 自动化日常重复性工作，将精力集中于创造性任务
- 构建数据采集 → 处理 → 输出的完整自动化管线
- 实现跨平台、跨系统的数据流转和业务协同
⚖️ 优点与不足
- +大幅减少重复性人工操作
- +可视化流程，清晰直观
- +可扩展性强，支持复杂场景
- −初始配置和调试需投入一定时间
- −强依赖外部服务的稳定性
- −复杂场景需具备一定技术基础
该工具使用 NOASSERTION 协议，商用场景请仔细阅读协议条款，必要时咨询法律意见。
AI Skill Hub 为第三方内容聚合平台，本页面信息基于公开数据整理，不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后，再部署至生产环境，并做好必要的安全评估。
📄 NOASSERTION — 请查阅原始协议条款了解具体使用限制。
🔗 相关工具推荐
❓ 常见问题 FAQ
AI Skill Hub 点评：SharpClaw 的核心功能完整，质量优秀。对于自动化工程师和运维人员来说，这是一个值得纳入个人工具库的选择。建议先在非生产环境试用，再逐步推广。
| 原始名称 | SharpClaw |
| 原始描述 | 开源AI工作流：Cross-platform AI chat and shell agent management application built with .NET 10。⭐8 · C# |
| Topics | C#AI跨平台 |
| GitHub | https://github.com/mkn8rn/SharpClaw |
| License | NOASSERTION |
| 语言 | C# |
收录时间：2026-05-16 · 更新时间：2026-05-30 · License：NOASSERTION · AI Skill Hub 不对第三方内容的准确性作法律背书。