starchild-ai-agent/@2004/skill-repo-publish

Skill Repo Publish

Push local SKILL.md changes to Starchild-ai-agent/official-skills.

Token

Always use GITHUB_TOKEN (not GITHUB_TOKEN_NEW — no org write access). Embed token in clone URL: https://${GITHUB_TOKEN}@github.com/Starchild-ai-agent/official-skills.git

Standard Workflow

# 1. Clone (always fresh /tmp dir)
git clone https://${GITHUB_TOKEN}@github.com/Starchild-ai-agent/official-skills.git /tmp/official-skills
cd /tmp/official-skills
git config user.name "nicholasDxy"
git config user.email "[email protected]"

# 2. Edit with Python replace (safer than sed)
with open("/tmp/official-skills/<skill>/SKILL.md", "r") as f:
    content = f.read()
content = content.replace("old text", "new text")
with open("/tmp/official-skills/<skill>/SKILL.md", "w") as f:
    f.write(content)

# 3. Verify diff before commit
git diff <skill>/SKILL.md

# 4. Commit + push
git add <skill>/SKILL.md
git commit -m "feat(<skill>): <summary>

<details>

Co-authored-by: Starchild <[email protected]>"
git push origin main

Version Bump Rules

| Change type | Bump | Example | |-------------|------|---------| | New capability or tool | minor | 3.2.0 → 3.3.0 | | Doc fix / wording | patch | 3.3.0 → 3.3.1 | | Breaking interface change | major | 3.x.x → 4.0.0 |

SKILL.md Writing Principles

Describe capabilities, don't enumerate dynamic data.

• ❌ Supported chains: ethereum, base, ...（list 16 chains）
• ✅ Supports all EVM chains via DeBank dynamic lookup (300+ chains)

DeBank chain lists, token lists, and other API-driven data change over time — never hardcode them. If there's a fallback map in code, describe its purpose ("offline map for common chains, not an exhaustive list"), don't copy it into the skill.

After Pushing

Always sync the local workspace skill too:

skill_manage(action="patch", name="<skill>", old_string="version: x.x.x", new_string="version: y.y.y")

So local skill stays in sync with the repo version.