Skip to content

README.en.md

Latest commit

 

History

History
118 lines (78 loc) · 6.06 KB

README.en.md

File metadata and controls

118 lines (78 loc) · 6.06 KB
Raw

English | 中文

GitHub Open Source Release Skill (github-open-source-release)

CI License: MIT Skill Standard Type: Documentation-Only

A standardized, safe, and compliant AI Agent skill specification for releasing local projects to GitHub as open-source repositories.

Distilled from real-world open-source engineering practices, this skill is tailored for modern AI agents such as Codex, Claude Code, and Oh My Pi. It enforces foundational principles: read-only audit first, strict pre-commit quality gates, Git author privacy protection, explicit user authorization, zero-credential leakage, and precise open-source license boundaries.

Release Workflow Demo (Offline Illustration)

GitHub Open Source Release Workflow Offline Illustration

Why This Skill?

When open-sourcing local projects to GitHub, developers and AI agents frequently encounter security, privacy, and compliance pitfalls:

  • Sensitive Data Leakage: Hardcoded API keys, private tokens, or developer local machine paths (e.g. C:/Users/...) left in code or documentation.
  • Privacy Breaches: Git commit history exposing enterprise employee IDs, internal domain emails, or personal private identifiers.
  • Accidental Artifacts: IDE settings (.idea, .vscode), debug logs, temporary caches, test coverage, or .env credential files mistakenly pushed to public repositories.
  • Suboptimal User Experience: Lack of bilingual README support, missing top-level language switcher links, broken external images, or fake single-frame GIF animations.
  • Unauthorized Actions: AI agents creating public remotes, pushing code, or deleting repositories without explicit user authorization.

This skill eliminates these risks through structured workflows and a Layered Verification Checklist.

Core Workflow Architecture

graph TD
    A[1. Read-Only Audit & Minimal Public File Set] --> B[2. Pre-Commit Gates: Test/Build/CI/Path Scan]
    B --> C[3. Git Status, Author Privacy & Clean Commit]
    C --> D[4. License Compliance & Dependency Boundaries]
    D --> E[5. Selectable Language README & Multi-Frame Demo]
    E --> F[6. Explicit User Authorization]
    F --> G[7. Repo Creation, Bilingual Metadata & Push]
    G --> H[8. Post-Push Remote Verification & Constraints]
Loading

Eight Core Specification Pillars

  1. Pre-Release Read-Only Audit & Minimal Public File Set: Complete read-only inspection, strictly excluding IDE configurations, local logs, and secrets while ensuring comprehensive .gitignore rules.
  2. Pre-Commit Quality Gates: Automated offline testing, build validation (when applicable), CI syntax verification, and deep credential/path scanning.
  3. Git Status & Author Privacy Audit: Verifying user.name and user.email to prevent enterprise ID or private email leakage, with semantic commit formatting.
  4. License Compliance & Dependency Boundaries: Distinguishing CLI invocations from source code distribution, adopting standard SPDX licenses (e.g., MIT/Apache-2.0).
  5. README Experience & Multi-Frame Media: User-selected primary language with matching secondary docs, top-level language switcher, and self-contained multi-frame demo animations.
  6. GitHub Repository Creation & Explicit Authorization: Requiring explicit user approval before performing externally visible operations; configuring bilingual descriptions and topic tags.
  7. Public Fork Cleanup & Private Reconstruction: Clarifying platform constraints and safely recreating private repositories with confirmed two-step authorization.
  8. GITHUB_TOKEN & Keyring Privilege Handling: Diagnosing environment variable overrides and scope deficiencies without leaking credentials.

Installation & Usage

1. Portable Installation to Your AI Agent

Copy the core specification file SKILL.md (or the entire directory) into your AI agent's skill path:

# Codex global skills directory
cp SKILL.md ~/.codex/skills/github-open-source-release/SKILL.md

# Claude Code / Oh My Pi skills directory
cp SKILL.md ~/.claude/skills/github-open-source-release/SKILL.md

2. Prompting the Agent

When ready to open-source a local project, instruct your agent:

Please use the github-open-source-release skill to safely and compliantly release this local project to GitHub.

The agent will load the specification and follow the 8-step workflow and verification checklist.

Project Nature (Documentation-Only Skill)

Important Notice: This repository is a Documentation-Only Skill specification. Its core deliverable is the standardized SKILL.md specification, bilingual documentation, and workflow assets.

Python packaging is not applicable to this repository. Quality and integrity are enforced via lightweight automated offline structural validation scripts and GitHub Actions CI rather than superficial packaging artifacts.

Local Structural & Security Validation

The repository includes a self-contained offline validation script covering front matter syntax, section completeness, relative media validity, animation frame counts, and secret/path scanning:

# Install lightweight validation dependencies
pip install pillow pyyaml

# Run local offline verification
python scripts/validate.py

Contributing & Security

  • Contributing Guide: See CONTRIBUTING.md for contribution rules and documentation sync standards.
  • Security Policy: See SECURITY.md for security principles and vulnerability reporting instructions.

License

This project is licensed under the MIT License. Copyright (c) 2026 divenire990.