Master the Full SkillHub Publishing Workflow: From Frontmatter to Approval
This guide walks you through every step of publishing a SkillHub skill—including completing required frontmatter metadata, adding an icon, choosing between CLI and web upload, passing the platform's security review, and avoiding common pitfalls—so your skill can go live on the first try.
In this article the author explains how to take a finished skill package and successfully publish it on SkillHub, treating the release process as a repeatable workflow rather than a one‑off task.
Pre‑publish: Complete the frontmatter
The frontmatter is a YAML header that acts as the skill’s identity card. The platform uses it to recognize the skill, its purpose, and compatible runtimes. Five fields are mandatory:
name : skill name, must match the folder name
description : a one‑sentence summary of what the skill does
version : semantic version, e.g., 1.0.0 category : the most appropriate category
platforms : list of supported platforms such as WorkBuddy, Claude Code, etc.
Example frontmatter from the author’s own skill:
---
name: wechat-article-writer-v2
description: "公众号文章自动生成技能(风云版v2)"
version: "2.0.0"
category: "content-creation"
platforms: [WorkBuddy,claude code,codex]
license: "MIT"
displayName:
en: "wechat article writer with windcloud"
zh: "风云的个人公众号文章生成器"
---The author missed the platforms field initially, causing the upload to fail; checking the table above prevents that mistake.
Icon: Default is fine, custom is better
If no icon is supplied, SkillHub uses a default one. Supplying a clear, function‑related icon improves credibility. The author simply cropped a square from a cover image and used it.
Publishing methods: CLI vs. web UI
CLI (for power users):
skillhub login # OAuth login, credentials saved to ~/.skillhub/auth.json
skillhub init --name wechat-article-writer-v2 --category content-creation
skillhub push
skillhub publishRunning these four commands packages the zip and pushes it to SkillHub.
Web UI (for those less comfortable with the terminal): log in at skillhub.cn, drag the zip onto the upload page, and let the platform auto‑populate the frontmatter form. An AI assistant can also generate missing metadata from a short description.
The author recommends the web UI for the first release to see the process, then switching to the CLI for frequent updates.
Review process: Not a formality
After submission the platform spends about five minutes reviewing the package. It checks:
Zip structure – SKILL.md must be at the root.
YAML parsing – any missing mandatory field leads to rejection.
Security scanning – content compliance, vulnerability scanning, and model‑safety assessment. The author cites an incident where a skill attempted to read ~/.ssh and was blocked.
SkillHub also applies a TRACE evaluation covering five dimensions: trustworthiness, reliability, applicability, compliance, and effectiveness. Failing the “T” (trust) dimension results in immediate elimination, similar to an exam’s pass/fail line.
Common pitfalls
Incorrect zip packaging : SKILL.md must be at the root; the zip size cannot exceed 10 MB.
Non‑incrementing version : the new version number must be higher than the current one; uploading 1.5.0 after 2.0.0 is rejected.
Visibility setting : four options exist – public, unlisted, private, org. To make the skill discoverable, choose public; otherwise it remains invisible to others.
Conclusion
Publishing a skill is the start of a new development cycle. After the skill is listed, user feedback and iterative updates follow, so the process of releasing is never truly finished.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
