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。
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。
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。
# 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
# 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"
Execution contracts for agent work.
Sykli is the contract layer between AI agents and the work they execute.
Agents can edit code, propose fixes, and adapt plans. They still need a reliable answer to basic questions:
Sykli answers those questions with typed execution contracts and structured evidence instead of terminal soup.
Agents should not regex CI logs. Sykli gives them typed failures, contract slices, retry hints, and evidence.
Sykli is an agent-readable execution layer for software work. It lets projects define build, test, deploy, review, and approval work as typed graphs; runs those graphs locally or in CI; and returns structured evidence that agents, reviewers, and downstream systems can trust.
Short version for the repo sidebar:
The contract layer between AI agents and the work they execute.// Content-addressed cache
s.Task("test").Run("go test ./...").Inputs("**/*.go", "go.mod")
// Containers and cache mounts
s.Task("build").
Container("golang:1.22").
Mount(s.Dir("."), "/src").
MountCache(s.Cache("go-mod"), "/go/pkg/mod").
Workdir("/src").
Run("go build -o app")
// Matrix expansion
s.Task("test").Run("go test ./...").Matrix("go", "1.21", "1.22", "1.23")
// Gates
s.Gate("approve-deploy").Message("Deploy?").Strategy("prompt")
s.Task("deploy").Run("./deploy.sh").After("approve-deploy")
// Artifact passing
s.Task("build").Run("go build -o /out/app").Output("binary", "/out/app")
s.Task("deploy").InputFrom("build", "binary", "/app/bin").Run("./deploy.sh /app/bin")
// Capability-based placement
s.Task("train").Requires("gpu").Run("python train.py")
// Conditional execution and secrets
s.Task("deploy").Run("./deploy.sh").When("branch == 'main'").Secret("DEPLOY_TOKEN")These examples use existing Go SDK APIs covered by SDK tests and conformance fixtures.
curl -fsSL https://raw.githubusercontent.com/false-systems/sykli/main/install.sh | bashOr download a binary for macOS or Linux.
<details> <summary>Build from source</summary>
git clone https://github.com/false-systems/sykli.git
cd sykli/core
mix deps.get
mix escript.build
sudo mv sykli /usr/local/bin/Requires Elixir 1.14+. </details>
| Use case | What Sykli gives you |
|---|---|
| Agentic workflows | Agents are executors; the graph defines what runs, what it depends on, and what evidence it produces |
| PR reviews | Experimental review nodes with constrained context and explicit primitive semantics |
| Release checks | SLSA v1.0 provenance attestations and structured run evidence |
| Security validation | Secret-scoped tasks, OIDC token exchange, and webhook hardening in the core engine |
| Human-in-the-loop approvals | Team Mode gates: an agent's run blocks on a gate, a reviewer approves from another machine, the heartbeat delivers the decision |
| Infrastructure validation | The same graph can target local execution, containers, Kubernetes, or a BEAM mesh |
| CI pipelines | The CI graph as code, with cache keys, dependency-level parallelism, and structured results |
sykli mcp exposes the execution graph and run evidence as tools an agent can call:
agent -> suggest_tests # what should I run for this change?
agent -> run_pipeline # run the graph or selected tasks
agent -> get_failure # what failed, semantically?
agent -> run_fix # correlate failure facts with source context
agent -> get_history # is this flaky or newly broken?That gives an AI coding agent a loop it can actually reason over:
change code -> ask what to run -> run it -> inspect typed failure -> repair -> rerunMCP is the transport. Sykli's value is the contract and evidence behind the tools: typed graph data, coded errors, failure semantics, agent hints, and contract slices.
| Language | Install | Default file |
|---|---|---|
| Go | go get github.com/false-systems/sykli/sdk/go@latest | sykli.go |
| Rust | sykli = "0.7.0" in Cargo.toml | sykli.rs |
| TypeScript | npm install sykli@0.7.0 | sykli.ts |
| Elixir | {:sykli_sdk, "~> 0.7.0"} in mix.exs | sykli.exs |
| Python | pip install sykli==0.7.0 | sykli.py |
The Go module is live today. Registry packages for the other SDKs are catching up to 0.7.0 — until your registry shows it, every SDK is usable directly from sdk/<lang>/ in this repository at the v0.7.0 tag.
All SDKs emit the same canonical contract shape.
高质量的AI工作流项目，值得关注
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。
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 |
| 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：2026-06-28 · License：MIT · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型，复制安装指令后粘贴到对应客户端