Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计提醒

integrate-skill综合技能

Agent Skill

integrate-skill 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。

总安装

282

周安装

12

GitHub Stars

124

下载量

99
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:integrate-skill(综合技能)
来源仓库:https://github.com/kenneth-liao/ai-launchpad-marketplace
仓库路径:skills/integrate-skill
安装命令:
npx skills add https://github.com/kenneth-liao/ai-launchpad-marketplace --skill integrate-skill
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/kenneth-liao/ai-launchpad-marketplace --skill integrate-skill

简介

用于查找、检索和筛选相关信息。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

  • 适合根据关键词或任务场景快速定位候选结果。
  • 通过 npx skills add 命令从指定仓库安装并使用。
  • 建议确认权限范围、维护状态及是否触发联网或文件读写操作。
  • integrate-skill 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Integrate Skill

Overview

This meta-skill formalizes the process of taking an existing skill from any source -- an external repository, a temp/ holding directory, a standalone file, or another project -- and decomposing it into the composable skill architecture. Source skills are rarely structured for composability. They bundle strategy, writing, platform knowledge, and workflow logic into a single file. This skill breaks that monolith apart, maps each piece to the right plugin and category, and delegates actual skill generation to skill-forge:create-skill.

Core Principle: Integration is decomposition. A source skill that "does everything" becomes multiple focused pieces that compose well. Nothing from the source gets dropped; everything gets placed.

When to Use

Use this skill when:

  • Migrating a skill from an external project into the marketplace
  • Importing a skill from the temp/ holding directory
  • Reproducing a skill from another Claude project or prompt collection
  • Converting a monolithic workflow into composable skills
  • Porting platform-specific automation into the architecture

Do NOT use this skill when:

  • Creating a skill from scratch (use skill-forge:create-skill directly)
  • The source is an idea or description, not an existing skill file

Integration Workflow

Execute all steps below in order. Do not skip steps. Steps marked MANDATORY must be completed before proceeding.

Step 1: Ingest the Source Skill

Read the source skill completely. Gather every file that constitutes the skill:

  • SKILL.md (or equivalent main file): The primary skill definition
  • Reference files: Any supporting documents, templates, or guidelines
  • Scripts or assets: Any code, images, or other bundled resources
  • Configuration: Any metadata, plugin manifests, or settings

Document the source location and file inventory:

Source: [path or repo URL]
Files found:
- [file 1]: [brief description]
- [file 2]: [brief description]
- ...

If the source skill has no clear boundary (e.g., it is embedded in a larger document), extract the relevant sections and note what was excluded and why.

Step 2: Analyze and Decompose

MANDATORY: Read skill-forge/skills/create-skill/references/taxonomy.md before proceeding.

For each distinct piece of content in the source skill, classify it using the taxonomy decision tree:

  1. Does it provide information that shapes other skills' output? --> Knowledge --> likely branding-kit/ or a domain-specific knowledge plugin
  2. Does it define how output sounds or feels? --> Personality --> likely writing/
  3. Does it do one specific thing well? --> Task --> likely writing/, content-strategy/, visual-design/, art/, research/, or personal-assistant/
  4. Does it sequence multiple task skills for a platform workflow? --> Orchestrator --> likely youtube/, substack/, or a new platform plugin
  5. Does it create or maintain other skills? --> Meta --> skill-forge/

For each piece, determine:

  • Category: Which of the five categories it belongs to
  • Plugin: Which existing plugin it maps to (or whether a new plugin is needed)
  • File type: SKILL.md (for new skills) or reference file (for supporting content)
  • Existing overlap: Whether an existing skill or reference already covers this content

Check for redundancy: Search existing plugins for skills and references that already cover the same ground. If overlap exists, the source content should extend the existing file rather than create a duplicate.

Flag potential loss: If any source content does not map cleanly to a category or plugin, flag it explicitly. Do not silently drop content.

Apply the split rule: if a source section is trying to be two things, split it. Strategy content and writing templates are almost always separate pieces even when the source bundles them together.

Step 3: Produce the Integration Design Doc

Create an integration design document at docs/plans/YYYY-MM-DD-<topic>-integration-design.md using today's date and a descriptive topic slug.

Use this template:

# Integration Design: [source skill name] -> [target plugin:skill]

## Summary
One paragraph describing what is being integrated and how the content splits.

## Architecture
ASCII diagram showing the decomposed structure — the resulting orchestrator
or primary skill and its delegations to task skills and references.

## Content Split
For each resulting piece:
### [N]. [Type]: `[plugin/path/to/file]`
What content goes here, why it belongs in this plugin.

## Files to Create/Modify
| Action | File | Purpose |
|--------|------|---------|
| Create | ... | ... |
| Update | ... | ... |
| Delete | ... | ... |

## Functionality Preservation
| Original Capability | Destination | Notes |
|---------------------|-------------|-------|
| ... | ... | ... |

## Design Decisions
Numbered list of key decisions with rationale.

## Patterns Followed
Bullet list referencing framework patterns.

Fill in every section. The Content Split section must account for every piece of source content identified in Step 2. The Functionality Preservation table must confirm that every capability of the source skill has a destination in the new architecture.

Step 4: Get User Approval

MANDATORY: Present the integration design doc to the user and wait for explicit approval before proceeding. This is a hard gate.

Present a summary of:

  • How many files will be created, modified, and deleted
  • Which plugins are affected
  • Any new plugins that need to be created
  • Any content flagged as potentially lost or ambiguous

Do NOT proceed to Step 5 until the user confirms the design. If the user requests changes, update the design doc and re-present.

Step 5: Execute the Integration

With the approved design, create each piece:

For each new skill identified in the design:

  1. Invoke skill-forge:create-skill with the classification, plugin placement, and content from the design doc
  2. Let skill-forge:create-skill handle template selection, composition hooks, validation, and invocation of authoring best practices (skill-creator:skill-creator, superpowers:writing-skills)
  3. Do NOT generate SKILL.md files directly -- always delegate to skill-forge:create-skill

For each reference file identified in the design:

  1. Create the reference file in the correct plugin's skills/[skill-name]/references/ directory
  2. Follow the existing reference file patterns in that plugin
  3. Keep reference files focused -- one topic per file, no fat reference files that dump everything into a single document

For new plugins (if the design requires them):

  1. Create the plugin directory structure: [plugin]/.claude-plugin/plugin.json, [plugin]/README.md
  2. Create the skill directories within the new plugin
  3. Follow existing plugin patterns for plugin.json and README.md

For all modified plugins:

  1. Bump the plugin version in .claude-plugin/plugin.json (increment the minor version)
  2. Update the plugin's README.md to document new skills or references

Step 6: Verify the Integration

Run verification against the approved design:

Framework validation:

  • Invoke skill-forge:create-skill Step 8 validation for each generated skill
  • Confirm every SKILL.md is under 500 lines
  • Confirm flat directory structure (max 2 levels)
  • Confirm frontmatter includes name and description

Functionality preservation:

  • Walk the Functionality Preservation table row by row
  • For each original capability, confirm the destination file exists and covers the capability
  • Flag any gaps

Composition validation:

  • Confirm generated task skills include voice and brand hooks where required
  • Confirm orchestrator skills delegate via plugin:skill syntax and do not generate content
  • Confirm reference files are loaded by the correct skills

Legacy syntax check:

  • Search generated files for outdated invocation patterns (bare skill names without plugin: prefix, old slash-command syntax)
  • Modernize any legacy patterns to use plugin:skill syntax

If any verification fails, fix the issue before proceeding.

Step 7: Clean Up

Propose cleanup of the source files:

  1. List all source files that were ingested in Step 1
  2. Present the list to the user and confirm deletion
  3. Delete confirmed source files (e.g., temp/ directories, imported files)
  4. Commit all changes with a descriptive message covering:

- The integration (new skills and references created) - The cleanup (source files removed)

Do NOT delete source files without user confirmation. If the user wants to keep the source files, skip deletion and note it.

Quality Checklist

Before finalizing the integration:

  • Source skill fully ingested (SKILL.md + all references/scripts/assets)
  • Every piece of source content classified using taxonomy decision tree
  • Content split mapped to existing plugins or new plugins identified
  • Integration design doc produced and user-approved
  • Functionality preservation table confirms zero loss of function
  • skill-forge:create-skill invoked for each resulting skill piece
  • Reference files created in correct foundation plugin directories
  • All generated skills pass skill-forge:create-skill Step 8 framework validation
  • Plugin versions bumped for all modified plugins
  • New plugins have complete structure (plugin.json, README.md) if created
  • Source cleanup proposed and user-confirmed
  • All changes committed

Common Pitfalls

  1. Losing functionality during decomposition: Splitting a source skill and quietly dropping pieces that do not fit neatly into a category. Every piece must appear in the Functionality Preservation table. If something does not map, flag it -- do not ignore it.
  2. Creating redundant skills: Not checking existing plugins before creating new skills or references. Always search for overlap in Step 2. If content-strategy:research already exists, add a reference file to it rather than creating a new research skill.
  3. Fat reference files: Dumping all source content into a single reference file. Reference files should be focused on one topic. Strategy content and writing templates are separate files even if the source bundled them together.
  4. Skipping create-skill: Generating SKILL.md files directly instead of invoking skill-forge:create-skill. The create-skill workflow enforces template selection, composition hooks, and validation. Bypassing it produces skills that miss framework requirements.
  5. Legacy syntax in generated files: Carrying over outdated invocation patterns from the source skill (bare skill names, slash commands, non-standard references). All generated content must use current plugin:skill syntax and follow current framework conventions.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude

32.35%
按下载量换算32

Codex

31.73%
按下载量换算31

Cursor

19.9%
按下载量换算20

Gemini CLI

8.35%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills