npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

vloop-skill

v0.2.0

Published

Three-layer closed-loop engineering skill for AI coding agents (plan-execute / acceptance-redesign / human-review). Installs into 40+ SKILL.md-compatible hosts: claude, codex, opencode, cursor, gemini, copilot, droid, zcode, antigravity, and more.

Readme

vloop — 三层闭环 Loop Engineering Skill

English | 中文

支持 12 种 agent CLI 的循环工程 skill:claude · codex · opencode · gemini · aider · copilot · cursor-agent · droid (Factory) · amp · qwen (通义千问 Code) · goose · kiro-cli —— 每个适配都经本机实测或官方文档核实,配置时做能力探测(flags 漂移快)。三层闭环:

L3 人类闭环      review 分支 / 改需求(PRD) / 打回     ← 唯一能改需求、批准不可逆动作的层
L2 产品验收环    异构只读 judge 逐条判验收标准;不过则重设计计划回灌 L1(≤3 轮)
L1 计划执行环    每轮新鲜上下文做一个任务 → 反压门(build/test/lint) → 绿则 commit 棘轮

核心机制:每层的"完成"只能由外层确认。verdict 是 schema 校验的文件而非魔法字符串;judge 与 executor 不同 backend 且物理只读;每层有硬上限(迭代/重设计轮次/预算/超时)+ 熔断器;人类门按动作类别(merge/deploy/publish/delete/charge/close)不可关闭。

目录

| 路径 | 内容 | |---|---| | docs/DESIGN.md | 三层闭环完整设计(架构、协议、配置器、适配层、安全) | | docs/RESEARCH.md | 调研摘要:Ralph 原典 / GitHub 6 实现 / HN 实战 / 本地实现拆解 / CLI 适配矩阵 | | skills/vloop/SKILL.md | Skill 入口(setup / run / resume / status / cancel 路由) | | skills/vloop/references/ | configurator(问答式配置器)· loop-protocol(三层执行手册)· adapters(多 backend 矩阵)· pitfalls(护栏检查单) | | skills/vloop/templates/ | loop.json · prd.json · plan.md · AGENT.md · pricing.json · 三个角色提示词模板 | | skills/vloop/scripts/ | vloop.sh(无人值守编排器)· adapter.sh(backend 调用 + 归一化 + 记账) |

安装

vloop 是标准 Agent Skill(SKILL.md 开放规范)—— 装一次,40+ 宿主通用:Claude Code、Codex、OpenCode、Cursor、Gemini CLI、GitHub Copilot、droid、goose、crush、amp、Kiro、ZCode、Antigravity、Trae、Windsurf……

# 推荐:一条命令,装进所有检测到的宿主
npx vloop-skill install

# 原理:
#   规范副本 -> ~/.agents/skills/vloop(codex/cursor/gemini/copilot/opencode/goose/crush/amp 原生读取)
#   + symlink -> ~/.claude/skills、~/.zcode/skills、~/.kiro/skills、~/.factory/skills、
#               ~/.gemini/antigravity/skills、~/.qwen/skills 等(只装检测到的宿主)

npx vloop-skill doctor       # 体检:依赖(jq/python3/git)、宿主、loop backend
npx vloop-skill uninstall    # 干净卸载(自带 manifest 追踪)

备选方式:

npx skills add wikieden/vloop        # 生态安装器(vercel-labs/skills,70+ agent)
npx github:wikieden/vloop install    # 直接从 GitHub 跑安装器,不依赖 npm 发布

无宿主纯 CLI 使用

模式 B 不需要任何 agent 宿主:

npx vloop-skill init   # 生成 .vloop/loop.json + prd.json 模板
# 编辑后:
npx vloop-skill run    # 无人值守三层循环;退出码 42 = 等人类 review

使用

/vloop setup            # 问答+选择题配置闭环(≤5 问/轮 ×2 轮)→ .vloop/loop.json + prd.json
/vloop run              # 模式 A:当前会话当编排器(可观察,首跑推荐)
/vloop run --unattended # 模式 B:vloop.sh 外部循环(过夜;退出码 42 = 等人类)
/vloop resume           # 人类 review 后恢复(PRD diff → 重生成计划)
/vloop status | cancel

依赖:jqpython3git,以及所选的 agent CLI。

已验证

  • 端到端 mock 测试(shim backend):L1 commit 棘轮/勾任务 → L2 只读 judge 判决提取 + passes 棘轮 → L3 暂停产物 + 退出码 42 ✓
  • 失败路径:gate 失败回滚、verdict 非法计失败迭代、熔断 trip1→replan、trip2→升级 L3 ✓
  • 跨 backend 成本台账(claude 原生 USD + gemini token 换算)✓
  • bash 3.2(macOS 默认)兼容 ✓

License

MIT