YouTip

A completed Skill needs to be packaged, tested, and documented before it can be distributed to team members or publicly released. This article introduces the complete process from development completion to release. * * * ## Pre-release Checklist Confirm each item before release to avoid distributing incomplete or defective Skills. | Check Item | Check Content | Reference Chapter | | --- | --- | --- | | Feature Complete | All planned features have been implemented and tested | Chapter 22 | | Error Handling | Common error scenarios have handling and prompts | Chapter 16 | | Accurate Description | Trigger test pass rate ≥ 85% | Chapter 24 | | Dependencies Declared | requirements.txt or compatibility field is complete | Chapter 20 | | No Hardcoded Keys | No sensitive information like API Keys, passwords in scripts | Chapter 28 | | Version Updated | version field in SKILL.md has been updated | Chapter 21 | | CHANGELOG Updated | This change has been recorded | Chapter 21 | | Debug Code Removed | No temporary print, debug checkpoints, TODO | Chapter 23 | * * * ## Packaging as .skill File Before release, the Skill directory needs to be packaged as a .skill file, which is the standard distributable format. ## Example # Basic packaging command python -m scripts.package_skill my-skill/ # Specify output path and version number python -m scripts.package_skill my-skill/ \ --output dist/my-skill-v1.2.0.skill \ --version 1.2.0 # Verify the packaged file is complete (list files in the package) python -m scripts.package_skill --list dist/my-skill-v1.2.0.skill Packaging complete: dist/my-skill-v1.2.0.skill (8 files, 42KB) contains: SKILL.md scripts/process.py scripts/requirements.txt assets/template.xlsx * * * ## Writing User Documentation At the end of SKILL.md or in a separate README.md, provide complete user instructions. # Data Analysis Assistant v1.2.0 ## What This Skill Can Do After uploading CSV or Excel files, automatically complete the following: - Count rows, calculate mean, max/min values, and null ratio for each column - Generate data quality report, marking columns that need cleaning - Output Excel report file with statistical charts ## Quick Start 1. Upload your data file (.csv or .xlsx) in the conversation 2. Say: "Help me analyze this data" or "Generate statistical report" 3. Wait for analysis to complete, download the generated Excel report ## System Requirements - Python 3.8+ - pandas >= 2.0 (automatically installed on first run) ## FAQ **Q: What if the file is too large?** Recommended file size is not more than 50MB. For larger files, filter or sample first before uploading. **Q: What encodings are supported?** Supports UTF-8 and GBK. If the file shows garbled characters, save as UTF-8 and re-upload. * * * ## Distribution Channels | Channel | Applicable Scenario | Operation Method | | --- | --- | --- | | Direct File Transfer | Small team internal sharing | Send .skill file, recipient uploads to Claude | | Git Repository | Team collaboration, needs version control | Commit to private/public repository, download from dist/ | | GitHub Releases | Official release, public sharing | Create Release, attach .skill file | | Internal CI/CD | Enterprise internal automated distribution | CI packages and pushes to internal file server | ### Publishing via GitHub Releases ## Example # Create Git tag and push git tag v1.2.0 -m"Release v1.2.0: New chart generation feature" git push origin v1.2.0 # Create Release with GitHub CLI and attach .skill file gh release create v1.2.0 \ dist/my-skill-v1.2.0.skill \ --title"My Skill v1.2.0" \ --notes"New chart generation feature, fixed memory issue with large file processing" * * * ## Installation Guide Page Template If your Skill is for multiple users, you can write a concise guide specifically for the installation steps. ## Installation Method 1. Go to (https://github.com/yourname/skills/releases/latest) 2. Download the `my-skill-v1.2.0.skill` file 3. Open Claude.ai → Settings → Skills 4. Click "Upload Skill", select the downloaded .skill file 5. After installation, start a new conversation to use it After installation, send the following in the conversation to verify successful installation: > "I uploaded a CSV file, please help me with data statistical analysis" If the Skill works normally, Claude will guide you through the analysis process according to the Skill's workflow.