mcpField
H
huawei-cloud-agentorchard-find-skills

huawei-cloud-agentorchard-find-skills

github.com/huaweicloudAgent workflows59206 installsUpdated 2d ago
Repository

About huawei-cloud-agentorchard-find-skills

This Agent skill searches AI Gallery for skills from natural-language requests and presents matching results. It can also direct users to skill details and subscription-based installation.

Key features

  1. 01Search skills in natural language
  2. 02Browse popular and available skills
  3. 03Open skill detail pages
  4. 04Guide subscription-based installation

Use cases

  1. 01Find an image classification skill
  2. 02Browse NLP skills in AI Gallery
  3. 03Look for data preprocessing skills

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/huaweicloud/huaweicloud-skills --skill huawei-cloud-agentorchard-find-skills

After installing, type/huawei-cloud-agentorchard-find-skillsin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

references/
acceptance-criteria.md
cli-installation-guide.md
dataflow-diagram.md
iam-policies.md
verification-method.md
scripts/
search_skills.py

Details

From the project README

Overview

This skill enables the Agent to efficiently search, discover, and install skills from AI Gallery. Simply input natural language; the Agent automatically understands intent and executes searches, returning results in natural language.

Scenarios

This skill enables the Agent to:

  • Search skills — Find skills by keyword
  • View skill details — Jump to AI Gallery detail page for full documentation
  • Subscribe to skills — Complete skill installation via AI Gallery subscription page
Usage Examples
  • "帮我找一个图像分类的 skill"
  • "AI Gallery 有哪些 NLP 相关的skill?"
  • "有没有数据预处理的 skill?"
  • "浏览所有可用的 AI Gallery skill"
  • "找一个文本生成的 skill"

Prerequisites

  • Agent runtime requires Python 3.6+ (script execution environment)
  • Network access to AI Gallery API (devdata.huaweicloud.com)
  • No authentication required — the API is public and read-only

Workflow

Step 1: Execute Search Script

Based on user input, run the search script (execute only once):

PYTHONIOENCODING=utf-8 python scripts/search_skills.py -k "<user input>"

Important

Agent must follow these rules:

  1. Execute only once — No repeated runs, no retries
  2. Output script results directly — No additional explanation, intermediate messages, or summaries
  3. No extra operations — Show exactly what the script outputs
  4. Use UTF-8 encoding — Must set PYTHONIOENCODING=utf-8 when running the script
  5. No direct API calls — Must use the script; do not bypass it to call the API directly

The script automatically handles all scenarios:

User Input TypeScript Output
Browse intent (e.g., "有什么skill")Hot skills + guidance
Search intent (e.g., "银行")Matching skill list
No results (e.g., gibberish)Guidance

The Agent only needs to display the script output directly to the user, without adding any extra operations.

Step 2: View Details & Subscribe (Optional)

After the user selects a skill, guide them to visit the detail page and subscribe.

Rules:

  • Only show the detail page URL when a specific show_id is available from the search results
  • Format: 详情页:https://pangu.huaweicloud.com/gallery/asset-detail.html?id={show_id}
  • Do NOT show URL templates or placeholder formats like ?id={show_id} to the user

The user clicks the "Subscribe" button on the detail page to complete installation.

Core Commands

CommandPurpose
PYTHONIOENCODING=utf-8 python scripts/search_skills.py -k "<keyword>"Search skills by keyword

Parameter Confirmation

ParameterRequiredDescriptionExample
-k, --keywordNoSearch keyword, supports space/comma/semicolon separated-k "银行"
--api-baseNoAI Gallery API base URL (has default)--api-base "https://..."

Reference Documents

  • references/cli-installation-guide.md — Environment setup and prerequisites
  • references/iam-policies.md — IAM policy notes (public API, no credentials needed)
  • references/verification-method.md — Functional verification methods
  • references/acceptance-criteria.md — Acceptance criteria and checklist
  • references/dataflow-diagram.md — Data flow diagram

Comments

Markdown supported