Token导航 LogoToken导航

Agent Skill · 研究检索

openspec-loop(openspec loop 搜索)

openspec-loop 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

openspec-loop 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 68,490。

总安装 233周安装 10GitHub Stars 1下载 82核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:openspec-loop(openspec loop 搜索)
来源仓库:https://github.com/viteinfinite/skills
仓库路径:skills/openspec-loop
安装命令:
npx skills add https://github.com/viteinfinite/skills --skill openspec-loop
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/viteinfinite/skills --skill openspec-loop

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

openspec-loop 用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词或任务场景快速定位候选结果。通过 npx skills add 命令从 viteinfinite/skills 仓库安装该技能。

安装前需确认权限范围、维护状态,以及是否会触发联网或文件操作。建议结合原始 README 继续核验具体循环逻辑和使用方式。

来源仓库内路径:skills/openspec-loop

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/viteinfinite/skills --skill openspec-loop;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

openspec-loop 怎么安装?

在终端执行 npx skills add https://github.com/viteinfinite/skills --skill openspec-loop。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

openspec-loop 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

openspec-loop 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

openspec-loop 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

openspec-loop 用的人多吗?

公开数据是累计安装 233 次,近一周安装 10 次,下载 82 次,GitHub 1 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

openspec-loop 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
openspec-loop

openspec loop 搜索 · 本页

2331需要联网未明确未通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

OpenSpec Loop: OpenSpec Subagent-Driven Development

Implement OpenSpec specs by dispatching fresh subagent per spec, with clean context isolation and sequential execution.

Core principle: Fresh subagent per spec + sequential execution + clean context = high quality isolated implementation

When to Use

digraph when_to_use {
    "Have OpenSpec specs to implement?" [shape=diamond];
    "Specs are independent changes?" [shape=diamond];
    "Want clean context per spec?" [shape=diamond];
    "openspec-subagent-driven-development" [shape=box];
    "Manual execution or other workflow" [shape=box];

    "Have OpenSpec specs to implement?" -> "Specs are independent changes?" [label="yes"];
    "Have OpenSpec specs to implement?" -> "Manual execution or other workflow" [label="no"];
    "Specs are independent changes?" -> "Want clean context per spec?" [label="yes"];
    "Specs are independent changes?" -> "Manual execution or other workflow" [label="no - tightly coupled"];
    "Want clean context per spec?" -> "openspec-subagent-driven-development" [label="yes"];
    "Want clean context per spec?" -> "Manual execution or other workflow" [label="no"];
}

Key characteristics:

  • Sequential execution (one spec at a time)
  • Fresh subagent per spec (no context pollution)
  • Each subagent starts with /clear
  • Spec archiving verifies completion

The Process

digraph process {
    rankdir=TB;

    "Run 'openspec view' to discover active specs" [shape=box];
    "Present active specs to user" [shape=box];
    "User confirms implementation order" [shape=box];
    "More specs remain?" [shape=diamond];

    subgraph cluster_per_spec {
        label="Per Spec";
        "Dispatch fresh general-purpose subagent" [shape=box];
        "Subagent executes /clear" [shape=box];
        "Subagent exports context usage" [shape=box];
        "Subagent executes /openspec:apply [spec-id]" [shape=box];
        "Subagent implements all spec changes using /openspec:apply [spec-id]" [shape=box];
        "Subagent archives the spec using /openspec:archive [spec-id]" [shape=box];
        "Verify spec archived successfully" [shape=box];
        "Report status to user" [shape=box];
    }

    "Provide implementation summary to user" [shape=box style=filled fillcolor=lightgreen];

    "Run 'openspec view' to discover active specs" -> "Present active specs to user";
    "Present active specs to user" -> "User confirms implementation order";
    "User confirms implementation order" -> "More specs remain?";
    "More specs remain?" -> "Dispatch fresh general-purpose subagent" [label="yes"];
    "Dispatch fresh general-purpose subagent" -> "Subagent executes /clear";
    "Subagent executes /clear" -> "Subagent exports context usage";
    "Subagent exports context usage" -> "Subagent executes /openspec:apply [spec-id]";
    "Subagent executes /openspec:apply [spec-id]" -> "Subagent implements all spec changes";
    "Subagent implements all spec changes" -> "Subagent archives the spec";
    "Subagent archives the spec" -> "Verify spec archived successfully";
    "Verify spec archived successfully" -> "Report status to user";
    "Report status to user" -> "More specs remain?";
    "More specs remain?" -> "Provide implementation summary to user" [label="no"];
}

Subagent Instructions Template

When launching a subagent, provide it with these exact instructions:

You are implementing a specific OpenSpec change proposal. Your workflow:

1. FIRST ACTION: Execute `/clear` to ensure clean context
2. THIRD ACTION: Execute `/openspec:apply [spec-id]` to apply the spec you're implementing
3. Implement all changes described in the spec
4. After successful implementation, archive the spec
5. Report completion and terminate

Spec to implement: [spec-details]

Example Workflow

You: I'm using OpenSpec Subagent-Driven Development to implement pending specs.

[Run 'openspec view' and parse active changes]
[Present list to user]

You: Found 3 active specs:
  1. add-base-image-uploads
  2. add-strapi-tag-creation
  3. some-other-feature

User: Implement in this order: add-base-image-uploads, then add-strapi-tag-creation

You: Confirmed. Starting implementation...

=== Spec 1: add-base-image-uploads ===

[Dispatch general-purpose subagent with spec details]
[Subagent executes /clear]
[Subagent exports context usage]
[Subagent executes /openspec:apply add-base-image-uploads]
[Subagent implements changes]
[Subagent archives spec]
[Verify archiving]

You: ✅ add-base-image-uploads completed and archived

=== Spec 2: add-strapi-tag-creation ===

[Dispatch fresh general-purpose subagent with spec details]
[Subagent executes /clear - completely fresh context]
[Subagent exports context usage]
[Subagent executes /openspec:apply add-strapi-tag-creation]
[Subagent implements changes]
[Subagent archives spec]
[Verify archiving]

You: ✅ add-strapi-tag-creation completed and archived

=== Summary ===

All requested specs implemented successfully:
  ✅ add-base-image-uploads
  ✅ add-strapi-tag-creation

Remaining: some-other-feature

Advantages

vs. Manual spec implementation:

  • Each spec gets dedicated, focused attention
  • Fresh context per spec (no confusion from previous work)
  • Automatic context usage tracking per spec
  • Spec archiving confirms completion

vs. Parallel implementation:

  • No context bleed between specs
  • Sequential execution prevents conflicts
  • Easier to debug if something goes wrong
  • Clear progress tracking

Quality benefits:

  • /clear ensures no mental contamination
  • Each subagent starts fresh
  • Failed specs don't affect subsequent specs
  • Clean audit trail per spec

Operational Constraints

NEVER:

  • Read the specs directly yourself - always delegate to subagents
  • Implement specs yourself - always delegate to subagents
  • Allow context to bleed between spec implementations
  • Skip the /clear action - it's critical for isolation
  • Proceed without verifying spec archiving
  • Implement specs in parallel - must be sequential
  • Use "superpowers:subagent-driven-development" skill

ALWAYS:

  • Verify spec exists and is valid before dispatching
  • Verify spec is archived after completion
  • Report failures and ask user how to proceed
  • Maintain clear status display

Error Handling

If subagent fails to implement its spec:

  • Report the failure to the user
  • Ask whether to: retry with new subagent, skip next spec, or abort
  • Document the failure reason

If spec cannot be applied:

  • Do not proceed to next spec without user confirmation
  • Get explicit direction before continuing

Progress Tracking

Maintain a clear status display showing:

  • Total specs to implement
  • Current spec being implemented
  • Completed specs
  • Remaining specs

Update after each spec completion.

Integration

Required workflow skills:

  • openspec:apply - Applies the spec to be implemented
  • openspec:archive - Archives completed specs

Related workflows:

  • superpowers:subagent-driven-development - Original task-based workflow (NOT for OpenSpec specs)

Termination

Terminate after:

  • All specs successfully implemented and archived
  • User explicitly requests termination
  • Unrecoverable error occurs and user chooses to abort

平台分布

Codex

33.99%
按下载量换算28

Claude

32.11%
按下载量换算26

Cursor

21.55%
按下载量换算18

Gemini CLI

9.15%
按下载量换算8

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills