经 AI Skill Hub 精选评估,逻辑MD 获评「推荐使用」。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 7.5 分,适合有一定技术背景的用户使用。
逻辑MD 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。
逻辑MD 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。
# 方式一:npm 全局安装 npm install -g logic-md # 方式二:npx 直接运行(无需安装) npx logic-md --help # 方式三:项目依赖安装 npm install logic-md # 方式四:从源码运行 git clone https://github.com/SingularityAI-Dev/logic-md cd logic-md npm install npm start
# 命令行使用
logic-md --help
# 基本用法
logic-md [options] <input>
# Node.js 代码中使用
const logic_md = require('logic-md');
const result = await logic_md.run(options);
console.log(result);
# logic-md 配置说明 # 查看配置选项 logic-md --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 LOGIC_MD_CONFIG="/path/to/config.yml"
The audit and governance layer for AI agent reasoning.
A portable, framework-agnostic file format for declaring agent reasoning as structured contracts: step DAGs, output schemas, tool permissions, and quality gates. Contracts are validated at compile time, executed deterministically, and produce auditable event traces by default. Where prose prompts give you behaviour, LOGIC.md gives you accountability.
Developed alongside and validated in production through Modular9, a visual node-based agent builder by the same author, where the contract-enforcement architecture was first proven at scale. Independently evaluated against Archon in a 60-trial controlled experiment (logic-md-archon-eval).
<p align="center"> <img src="docs/assets/hero-layers.svg" alt="LOGIC.md sits between agent identity (CLAUDE.md), capability (SKILL.md), and protocols (MCP, A2A) as the missing declarative reasoning layer." width="100%"/> </p>
---
A LOGIC.md file is a markdown document with a YAML frontmatter block. The YAML contains the machine-parseable spec. The markdown body contains human-readable design rationale. Two fields are required (spec_version and name). Everything else is optional.
Twelve sections: imports, reasoning, steps, contracts, quality_gates, decision_trees, fallback, global, nodes, edges, visual, and metadata fields.
Imports: compose external LOGIC.md files with namespace merging and circular detection. Build libraries of reusable reasoning patterns.
Expression engine: {{ ... }} expressions with dot notation, comparisons, logical ops, array methods (.length, .avg, .min, .max, .every, .some, .contains), and ternary expressions. Zero eval, zero Function constructor.
DAG resolver: Kahn's algorithm for topological sort, cycle detection, and parallel execution level computation. Steps with no dependency relationship execute in parallel automatically.
See docs/SPEC.md for the full v1.0 specification.
---
logic-md compile my-agent.logic.md --step gather_sources
Install globally:
npm install -g @logic-md/cli
Use the 9-command interface:
```bash
@logic-md/core on npmsdks/python/| Package | Description | Install |
|---|---|---|
@logic-md/core | Parser, validator, expression engine, DAG resolver, reasoning compiler | npm i @logic-md/core |
@logic-md/cli | 9 commands, 16 templates, shell completion for bash/zsh/fish | npm i -g @logic-md/cli |
@logic-md/mcp | 7 MCP tools over stdio and HTTP, works with any MCP host | npm i @logic-md/mcp |
---
Five slash commands via the built-in integration:
/logic:status : detect and show LOGIC.md files in your project
/logic:apply : load a spec and apply its reasoning scaffold to the current task
/logic:validate : validate all LOGIC.md files in the project
/logic:init : scaffold a new LOGIC.md interactively
/logic:compile : show the compiled prompt segment for a specific step
Four reasoning workflow templates ship with the plugin: code-review, debug-workflow, refactor, architecture.
---
validate, lint, compile, init, test, watch, fmt, diff, completion| Handles | Portable | |
|---|---|---|
| CLAUDE.md / AGENTS.md | Identity, project context, code style | Markdown |
| OpenClaw SOUL.md | Personality, behavioural rules | Markdown |
Cursor .mdc rules | Coding conventions with activation modes | Markdown |
| MCP | Agent ↔ tool connectivity | Protocol |
| A2A Protocol | Agent ↔ agent communication | Protocol |
| BAML | Typed output schemas, cross-language codegen | Custom DSL |
| Instructor | Pydantic validation for individual LLM calls | Library |
| LangGraph | Reasoning as imperative StateGraph code | Python |
| CrewAI | YAML for roles/tasks, reasoning as a bool | Partial |
| AutoGen | Conversation patterns, reasoning in messages | Python |
| DSPy | Composable signatures + prompt optimization | Python |
| **LOGIC.md** | **Step DAGs, contracts, quality gates, multi-agent** | **Markdown / YAML** |
---
逻辑MD是一个有前景的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。
AI Skill Hub 点评:逻辑MD 的核心功能完整,质量良好。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 | logic-md |
| 原始描述 | 开源AI工作流:The declarative reasoning layer for AI agents. A portable, framework-agnostic f。⭐22 · TypeScript |
| Topics | AI工作流TypeScript |
| GitHub | https://github.com/SingularityAI-Dev/logic-md |
| License | MIT |
| 语言 | TypeScript |
收录时间:2026-06-09 · 更新时间:2026-06-09 · 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