Agent Skills › espressif/esp-claw › skills_lab_search

skills_lab_search

GitHub

用于在ESP-Claw技能中心搜索和发现Skills及Apps，支持关键词、标签、外设及分类过滤，并引导用户下载选定项目。

components/claw_capabilities/cap_skill_mgr/skills/skills_lab_search/SKILL.md espressif/esp-claw

触发场景

用户希望搜索或发现ESP-Claw技能中心的Skills和Apps 用户提供主题、功能、外设、标签或类别以查找匹配的技能或应用用户不知道安装所需的精确包ID和类型

安装

npx skills add espressif/esp-claw --skill skills_lab_search -g -y

更多选项

非标准路径

npx skills add https://github.com/espressif/esp-claw/tree/master/components/claw_capabilities/cap_skill_mgr/skills/skills_lab_search -g -y

不安装直接使用

npx skills use espressif/esp-claw@skills_lab_search

指定 Agent (Claude Code)

npx skills add espressif/esp-claw --skill skills_lab_search -a claude-code -g -y

安装 repo 全部 skill

npx skills add espressif/esp-claw --all -g -y

预览 repo 内 skill

npx skills add espressif/esp-claw --list

SKILL.md

Frontmatter

{
    "name": "skills_lab_search",
    "metadata": {
        "cap_groups": [
            "cap_http_request",
            "cap_boards"
        ]
    },
    "description": "Search Skills and Apps from the ESP-Skill hub by keywords and advanced filters, then guide the user to download a selected Skill or App with skills_lab_downloader."
}

Skills Lab Search

Use this skill when the user wants to search or discover Skills and Apps from the ESP-Claw's Skill hub.

When to use

The user wants to search for Skills and Apps from the ESP-Claw's Skill hub.
The user provides a topic, feature, peripheral, tag, or category and wants matching Skill or App candidates.
The user does not know the exact package id and kind needed by skills_lab_downloader.

Hard rules

Before searching, make sure cap_http_request is enabled and available. If it is unavailable, stop and tell the user to enable HTTP requests first.
Prefer English keywords for better search results unless the user provides a specific query.
Always URL encode the full query string before appending it to https://skills-lab.esp-claw.com/api/search?q=.
Use the search API endpoint directly. Do not scrape the web UI page for results.
If the user asks which tags, peripherals, or categories are available, fetch https://skills-lab.esp-claw.com/raw/tags.json and summarize the relevant values.
Preserve advanced search syntax exactly before URL encoding: t:"tag", p:"peripheral", and c:"category".
Use featured=true when the user wants to filter results to featured packages only.
If the user asks to list all available packages, request featured packages only with featured=true and no q parameter, present those featured packages, and ask the user for more specific search criteria.
After presenting results, remind the user that installation requires the exact package id and kind and can be done with skills_lab_downloader.

Package selection

Search results may contain Skills and Apps. Preserve each result's kind (skill or app) and id; display the type alongside the title. If the user requests only one type, filter results by kind. Never guess a missing type.
When handing a selection to skills_lab_downloader, pass kind and id unchanged. App and Skill ids can overlap.

Search syntax

Free text keywords: bilibili fans
Tag filter: t:"info"
Peripheral filter: p:"display"
Category filter: c:"utility"
Filters can be combined with free text keywords: bilibili fans t:"info"
A filter-only query is valid: t:"info"
Featured-only filter: featured=true

Available tag, peripheral, and category values can be checked at:

https://skills-lab.esp-claw.com/raw/tags.json

Workflow

If the user asks to list all available packages, request featured packages only with no q parameter:

https://skills-lab.esp-claw.com/api/search?featured=true

Present the featured package results, then ask the user for more specific search criteria before doing a broader search.
Ask for search keywords if the user did not provide any searchable terms, filters, or a request for featured packages.
Build a search query with English keywords when possible.
Include advanced filters only when the user asks for them or when they are clearly implied.
Add featured=true when the user wants featured packages only.
URL encode the complete q query value when q is used.
Request the search API:

https://skills-lab.esp-claw.com/api/search?q=<url_encoded_query>

When using both q and featured=true, request:

https://skills-lab.esp-claw.com/api/search?q=<url_encoded_query>&featured=true

Read the returned search results and present the best matches with exact ids, kinds, titles, and short descriptions when available.
If there are no results, suggest a broader query, fewer filters, or checking available values from tags.json.
If the user chooses a Skill or App to install, activate skills_lab_downloader and pass the exact kind and id from the result.

Examples

Query: bilibili fans

https://skills-lab.esp-claw.com/api/search?q=bilibili+fans

Query: bilibili fans t:"info"

https://skills-lab.esp-claw.com/api/search?q=bilibili+fans+t%3A%22info%22

Query: t:"info"

https://skills-lab.esp-claw.com/api/search?q=t%3A%22info%22

Query: t:"info" quote

https://skills-lab.esp-claw.com/api/search?q=t%3A%22info%22+quote

Featured-only query: bilibili fans

https://skills-lab.esp-claw.com/api/search?q=bilibili+fans&featured=true

List featured packages:

https://skills-lab.esp-claw.com/api/search?featured=true

Notes

Search results are only candidates. Do not install a result until the user selects the exact package id and kind.
Do not guess a package id from a visible title. Use the exact package id and kind returned by the search API.
skills_lab_downloader handles metadata checks, peripheral compatibility, and installation.