AI工作流
AI Skill Hub 强烈推荐：AI工作流 是一款优质的Agent工作流。AI 综合评分 8.0 分，在同类工具中表现稳健。如果你正在寻找可靠的Agent工作流解决方案，这是一个值得深入了解的选择。
📚 深度解析
AI工作流 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
AI工作流 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
AI工作流 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/false-systems/sykli cd sykli # 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 sykli --help # 基本运行 sykli [options] <input> # 详细使用说明请查阅文档 # https://github.com/false-systems/sykli
# sykli 配置说明 # 查看配置选项 sykli --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 SYKLI_CONFIG="/path/to/config.yml"
简介
sykli
<p align="center"><strong>Know what actually ran. Not what someone says ran.</strong></p>
<p align="center"> <img alt="License: MIT" src="https://img.shields.io/badge/license-MIT-blue"> <img alt="Rust 1.85+" src="https://img.shields.io/badge/rust-1.85%2B-orange"> <img alt="Linux, macOS, Windows" src="https://img.shields.io/badge/platforms-linux%20%7C%20macos%20%7C%20windows-lightgrey"> <img alt="No server" src="https://img.shields.io/badge/runs-locally%2C%20no%20server-success"> </p>
sykli is a content-addressed evaluator for declared work: a small command-line tool that runs the commands you name in a repository and writes a receipt binding each result to the exact files, contract and command it came from. Anyone, human or agent, can later ask whether a receipt is still true for the code in front of them. It is available as a single binary for Linux, macOS and Windows.
sykli is not a CI service, a task tracker or an agent runner. It sits under the tools you already use (cargo, npm, go, your CI runner) and in front of the people and agents who need to trust their results without re-doing them or taking a summary on faith.
tasks · commands · input files"] target["sykli.production.json
source · build · required checks"] rules["requirements
for a pull request"] end subgraph engine["sykli runs and records"] run["run / plan / verify"] produce["produce / resume"] inspect["inspect / assess"] end subgraph evidence["Evidence, content-addressed"] receipt["receipt
what ran, on which tree"] artifact["artifact + checks
resumable by ID"] verdict["evidence bundle
established · refuted · unproven"] end contract --> run --> receipt target --> produce --> artifact rules --> inspect --> verdict receipt -. "exit codes + JSON" .-> readers["humans · scripts · agents"] artifact -.-> readers verdict -.-> readers
Features
- Content-addressed everything. Contracts, receipts, inputs, artifacts and saved evidence are named by the hash of their bytes. One byte different is a different identity and a different result. No "latest", no timestamps to trust. - Delta planning and caching. sykli plan --changed PATH names the tasks a change requires; unchanged work is served from cache, and a task never inherits a pass recorded against inputs it did not see. - Resumable typed production. Capture source, build an artifact, run required checks, stop at any boundary; another worker resumes by ID from the saved records, without the previous worker's conversation. Linux and macOS. - Pull-request evidence. Read a PR's CI runs and reviews through your gh login into an immutable bundle and assess your own declared requirements against the exact head commit, offline and reproducibly. Advisory: nothing is merged, triggered, posted or certified. - One interface for humans, scripts and agents. Readable text on the screen, one versioned sykli-*.v1 JSON document on stdout with --json, and fixed exit codes for every command. - Honest about limits. A receipt says what ran, never what it meant. Every assessment prints its trust boundary. verify proves consistency with the tree and contract, not who wrote the receipt.
Runtime requirements
- git, for the tree identity every receipt is bound to. - A POSIX sh on PATH to run tasks; on Windows, Git for Windows provides one. - For typed production: Linux or macOS, and the toolchain your target uses. - For pull-request evidence: the GitHub CLI (gh), logged in. sykli never stores credentials.
Getting started
- Install the binary, then in any Cargo, npm or Go repository: sykli init && sykli run. - Read the thirty-second walkthrough below to see what a receipt is. - Three situations sykli is built for: run checks and get a receipt, build, stop, let someone else finish, ask whether a pull request is ready. - Driving sykli from an agent: docs/agents.md. Every command documents its flags and exit codes under --help.
2. Build something, stop, let someone else finish
Declare a target: the source files, the build, the checks that must pass. sykli captures the source, builds, runs the checks, and saves the artifact with its identity. Stop early and another terminal or agent resumes with one ID.
sykli init --production --smoke '"$SYKLI_INPUT_executable" --help'
sykli produce sykli --stop-after build # exits 1: built, checks remain
sykli resume PRODUCTION_ID # someone else finishes the checksThe next worker needs the ID and the local store, not a conversation. Edit the source and you get a new production; old passing checks never count for new bytes. A lost attempt can be abandoned explicitly and retried; nothing is ever marked done that did not run.
Install
curl -fsSLO https://raw.githubusercontent.com/false-systems/sykli/main/install.sh
sh install.sh v0.6.0Or cargo install --git https://github.com/false-systems/sykli --tag v0.6.0 --locked sykli with Rust 1.85 or newer. Each release carries tarballs for Linux and macOS (x86_64 and aarch64), a Windows x86_64 zip, a Homebrew formula and SHA256SUMS. action.yml in this repository is a GitHub Action that installs the release matching its ref and runs the graph.
Releases and API stability
Every machine-readable document carries a versioned schema (sykli-contract.v1, sykli-receipt.v1, sykli-assessment.v1, …). A schema's meaning never changes once published; additions are optional fields, and anything incompatible is a new version. Exit codes are part of the interface and are listed in each command's --help. Releases are tagged vX.Y.Z and built by the release workflow from that tag; see the changelog.
高质量的AI工作流项目，值得关注
- 需要让 Claude / Cursor 操作本地工具的 AI 工程师
- 构建多智能体协作系统的 Agent 开发者
- 配置 MCP 服务器时建议使用 stdio 传输 + JSON-RPC，避免暴露公网
- 生产部署优先使用 Docker Compose 隔离依赖，并挂载 volume 持久化数据
- 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）
- MCP 配置路径拼错或权限不足，重启 Claude Desktop 才生效
- 容器内无法访问宿主机 localhost — 使用 host.docker.internal
- Docker：sykli 提供官方镜像，docker compose up 一键启动
- CLI：直接 npm install -g / pip install，命令行调用
- 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
- 需要让 Claude / Cursor 操作本地工具的 AI 工程师
- 构建多智能体协作系统的 Agent 开发者
- 配置 MCP 服务器时建议使用 stdio 传输 + JSON-RPC，避免暴露公网
- 生产部署优先使用 Docker Compose 隔离依赖，并挂载 volume 持久化数据
- 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）
- MCP 配置路径拼错或权限不足，重启 Claude Desktop 才生效
- 容器内无法访问宿主机 localhost — 使用 host.docker.internal
👥 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
- +MIT 协议，可免费商用
- +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
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。
✅ MIT 协议 — 最宽松的开源协议之一，可自由商用、修改、分发，仅需保留版权声明。
🔗 相关工具推荐
❓ 常见问题 FAQ
总体来看，AI工作流 是一款质量优秀的Agent工作流，在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态，建议收藏备用，结合自身场景选择合适时机引入使用。
| 原始名称 | sykli |
| 原始描述 | 开源AI工作流：The contract layer between AI agents and the work they execute.。⭐9 · Elixir |
| Topics | AI工作流Elixir |
| GitHub | https://github.com/false-systems/sykli |
| License | MIT |
| 语言 | Elixir |
收录时间：2026-06-28 · 更新时间：2026-07-03 · License：MIT · AI Skill Hub 不对第三方内容的准确性作法律背书。