← Back to Skills Marketplace
Milestone Execution
by
Siesta-0402
· GitHub ↗
· v3.0.1
· MIT-0
132
Downloads
1
Stars
0
Active Installs
8
Versions
Install in OpenClaw
/install milestone-execution
Description
多阶段任务执行器 - 用户控制节奏，AI在独立工作会话中逐阶段执行任务。每个 milestone 完成后暂停汇报，用户确认后再继续下一阶段。
Usage Guidance
该技能总体自洽且实现简单，但在安装前请注意：
- 它会在你的工作目录写入并更新 .milestone-state.json（用于崩溃恢复和进度跟踪）；如果你敏感于本地文件，请在受控/沙盒目录中运行或先备份。
- SKILL.md 明确允许“工作会话调用工具、修改文件等”，所以当你按“继续”或允许工作会话运行时，该会话可能会运行命令或修改项目文件。确保只在信任的环境或非关键代码库中使用，或先审查并测试 state.js 等脚本。
- 没有远程网络调用或凭据请求，这降低了外部数据泄露风险；若未来版本引入网络功能，应再次审核安装/网络端点和所请求的凭据。
总体建议：如果功能符合你的需求，可在受控目录中试用；若对自动执行命令或文件更改敏感，请先审查/修改工作会话的执行细节或限制其运行范围。
Capability Analysis
Type: OpenClaw Skill
Name: milestone-execution
Version: 3.0.1
The skill bundle implements a multi-stage task orchestration framework called 'Milestone Execution'. It uses OpenClaw's session management primitives (sessions_spawn, sessions_send, sessions_yield) to allow an AI agent to execute complex tasks in a controlled, step-by-step manner. The provided scripts (state.js) and instructions (SKILL.md) focus on state persistence via a local JSON file and user-driven progress. There are no indicators of data exfiltration, malicious persistence, or harmful prompt injection; the code and instructions are well-aligned with the stated purpose of task management.
Capability Assessment
Purpose & Capability
技能名与描述（多阶段/里程碑执行器）与所包含的文件和脚本一致。唯一的功能需求是在本地维护 .milestone-state.json 并在工作会话中逐步执行 milestone，未请求与该目的不相干的凭据、二进制或配置路径。
Instruction Scope
SKILL.md 明确描述了工作会话的生命周期和通信（sessions_send / sessions_yield 等），并允许工作会话“调用工具、修改文件等”。这是该技能的预期行为，但也意味着工作会话在执行里程碑时可能运行命令或修改工作目录下的文件——用户应理解这会授予技能在其运行目录内修改文件的能力。
Install Mechanism
无安装规范（instruction-only）且没有远程下载或可执行安装步骤，随附的脚本为小型本地脚本（start.js 已弃用，state.js 管理状态），风险低。
Credentials
不要求任何环境变量、凭据或外部配置路径。所有状态和操作均由本地状态文件与平台会话 API 管理，所需权限与技能目的相称。
Persistence & Privilege
flags 显示 always: false，技能为用户可调用且允许模型调用（平台默认）。不会强制常驻，也不试图修改其他技能或系统级配置。唯一的持久化是写入工作目录的 .milestone-state.json（用于断点恢复），这是功能所需且可见。
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install milestone-execution - After installation, invoke the skill by name or use
/milestone-execution - Provide required inputs per the skill's parameter spec and get structured output
Version History
v3.0.1
添加 README.md 功能说明
v3.0.0
重写为单一工作会话架构，修复了每阶段新建会话的问题
v2.0.4
v2.0.4: Genericize SKILL.md - remove project-specific examples
v2.0.3
v2.0.3: Generic README without project-specific details
v2.0.2
v2.0.2: Update README with v2.0 features
v2.0.1
v2.0.1: Rich reporting, progress tracking, time estimates, error recovery
v2.0.0
v2.0: Rich reporting, progress tracking, time estimates, error recovery (retry/skip/rollback), parallel milestones, context persistence, history record, automatic cleanup
v1.0.0
Initial release
Metadata
Frequently Asked Questions
What is Milestone Execution?
多阶段任务执行器 - 用户控制节奏，AI在独立工作会话中逐阶段执行任务。每个 milestone 完成后暂停汇报，用户确认后再继续下一阶段。 It is an AI Agent Skill for Claude Code / OpenClaw, with 132 downloads so far.
How do I install Milestone Execution?
Run "/install milestone-execution" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Milestone Execution free?
Yes, Milestone Execution is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Milestone Execution support?
Milestone Execution is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Milestone Execution?
It is built and maintained by Siesta-0402 (@siesta-0402); the current version is v3.0.1.
More Skills