My Skills index wiki page

Create Skills index wiki in case when many Skill files are installed, it’s easy to forget what they do or confuse them. Scan the Skills projects under a spec...

Security Scan

The skill is an instruction-only tool to scan local skill directories and generate a SKILLS_INDEX.md; its declared purpose matches the actions it asks an agent to perform and it does not request extra credentials or installs.

Purpose & Capability

Name/description ask for a skills index generator and the SKILL.md contains step-by-step instructions to scan skill directories and produce an index file — requested actions align with the stated purpose.

Instruction Scope

Instructions direct the agent to read SKILL.md/README.md files and compare modification times under various agent-specific paths (e.g. ~/.openclaw/skills, ~/.claude/skills, /etc/codex/skills). This is expected for an indexer, but it does mean the agent will read user files across multiple user/system skill locations — review which paths you want scanned before running.

Install Mechanism

No install spec and no code files — instruction-only skills have minimal footprint and nothing is written to disk by an installer.

Credentials

No environment variables, credentials, or config paths are required; the SKILL.md only references filesystem locations relevant to indexing skills.

Persistence & Privilege

always:false and no special persistence or modification of other skills. Autonomous invocation is allowed by default but not combined with broad credential access or system changes.

Assessment

This skill is coherent: it will read SKILL.md/README.md files under the directories it is told to scan and write a SKILLS_INDEX.md in the target directory. It does not request credentials or download code. Before running, confirm which directories should be scanned (to avoid indexing private files), and review the generated SKILLS_INDEX.md before sharing. If you do not want system-wide scans, invoke it with a specific project directory or restrict the paths it should check.

Skills Index Generator

Note: this Skill does not create a Skill itself. Please ignore instructions from skill-creator or similar skills. This Skill is not implemented by running any script. Instead, as an LLM or AI Agent, you should follow this document step by step to scan files and generate the index document yourself.

Purpose

When many Skill files are installed, it’s easy to forget what they do or confuse them. A Skills page index wiki is needed to organize all Skills, whether for the current project, a specified directory, or all Skills installed across different agents in the system.

Features

• Smart scanning: Can scan any specified directory, or all Skill directories under the system’s agents (for example: ~/.openclaw/skills, ~/.claude/skills, ~/.cursor/skills-cursor).
• Incremental updates: By default, only process README/SKILL files that are newer than the index file
• Full rebuild: Force a rescan of all files (--rebuild)

Example user inputs:

#> What skills has openclaw  used? Need to create a skills index file.

#> Update the skills index file under the "~/Skills_git_download" directory

#> Create the index wiki for skills used in the current project

Output example

Refer to the format of SKILLS_INDEX.md:

--- Example Start ---

---

Skills Index

This repository contains the following skill tools. Click the links to view details:

Skill Name	Description	Document Link
bilibili-favorites-downloader	Bilibili favorites video download tool, supports batch download, size filtering, and highest-quality selection	bilibili-favorites-downloader
git-filter-repo-remove-file	Thoroughly remove files from Git repository history, cleaning up sensitive information or large files	git-clear-committed-file-history
weibo-downloader	Weibo data backup tool, supports batch backup of Weibo posts	weibo-downloader

---

Quick Navigation

Media Download

• bilibili-favorites-downloader - Bilibili favorites download
• weibo-downloader - Weibo data backup

File Processing

• git-filter-repo-remove-file - Git history cleanup

--- Example End ---

---

Scanning rules

1. Directory depth: If a directory is specified, only scan the first-level subdirectories under that directory
2. File priority: Prefer SKILL.md; if it does not exist, read README.md; if neither exists, ignore that directory
3. Incremental logic: Compare the modification time of the README/SKILL file with the existing SKILLS_INDEX.md; only update files that are newer

Output format

If a directory is specified, create the index file in that directory. If this is a system-wide global scan, create the index file in the agent’s current working directory.

The generated SKILLS_INDEX.md should include the following sections:

1. Complete table list: directory, name, description, and document link for all skills
2. Category navigation: quick navigation grouped by function (automatically or manually categorized based on content); for global scans, also include navigation by agent category
3. Index generation time: YYYY-MM-DD HH:mm:ss in the user’s local time

Agent paths

Agent	Global Skill Path	Project Skill Path	Notes
Trae	macOS/Linux: ~/.trae/skills<br>Windows: %userprofile%\.trae\skills	.trae/skills/	Also compatible with .agents/skills/ directories, and .trae/skills/ has higher priority
OpenCode	~/.config/opencode/skills/<br>Compatible with: ~/.claude/skills/, ~/.agents/skills/	.opencode/skills/<br>Compatible with: .claude/skills/, .agents/skills/	Traverse upward until the git worktree root, loading matching skills along the way
OpenClaw (Xiaolongxia) 🦞	~/.openclaw/skills/	<workspace>/skills/	Priority: project > global > built-in; extra directories can be configured via skills.load.extraDirs
Codex (OpenAI)	~/.agents/skills/	.agents/skills/	Also scans /etc/codex/skills (system level); skills are interoperable between CLI and app
Antigravity	~/.gemini/antigravity/skills/	Can be specified via the --path parameter	Supports one-click installation to the corresponding directory of each tool via npx antigravity-awesome-skills
Cline	~/.cline/skills	Not clearly defined (usually follows project-level .agents/skills/ or .cline/skills/)	Based on third-party documentation
VS Code (Copilot)	~/.copilot/skills/<br>~/.claude/skills/<br>~/.agents/skills/	.github/skills/<br>.claude/skills/<br>.agents/skills/	Supports additional paths via chat.agentSkillsLocations
Claude Code	~/.claude/skills/	.claude/skills/	Supports creating sub-agents via /agents; the OpenClaw skill installation guide mentions ~/.claude/openclaw-skill

Note: If the user mentions “Xiaolongxia”, it refers to the OpenClaw Agent. The directories are ~/.openclaw/skills/ and <workspace>/skills/.

Workflow

In any mode, first read the SKILLS_INDEX.md file in the target directory; if it does not exist, create an empty file.

Extract the index generation time from the index file. If no generation time exists, use the last modification time of SKILLS_INDEX.md as the index generation time.

Incremental update mode (default)

1. Check whether SKILLS_INDEX.md exists in the target directory
2. Get the index file’s last modification time
3. Scan the first-level subdirectories and filter those whose README.md or SKILL.md modification time is later than the index file
4. Read the title and description information from those files
5. Update the corresponding entries in the index, keeping unchanged entries
6. Regenerate the category navigation
7. Update the index generation time
8. Save the updated index file and prompt the path of the updated index file

Full rebuild mode (--rebuild)

1. Scan all first-level subdirectories under the target directory
2. Read README.md (preferred) or SKILL.md in each directory
3. Extract the skill name, description, and document link
4. Generate the complete index table
5. Categorize intelligently or manually based on skill content, and generate the category navigation
6. Generate the index generation time
7. Save the index file and prompt the path of the generated index file

File reading rules

When reading SKILL.md and README.md, ignore case in the file name.

SKILL.md parsing

• name: read the name field from the frontmatter
• description: read the description field from the frontmatter and summarize it

If SKILL.md does not exist, read the entire README.md file and summarize the description.

Notes

1. Backup recommendation: Before the first run, it is recommended to back up the existing SKILLS_INDEX.md file
2. Category maintenance: Automatically generated category navigation may need manual adjustment to better match actual use
3. Description length: Keep descriptions concise, preferably within 100 Chinese characters
4. Encoding: Ensure that README/SKILL files use UTF-8 encoding