fix(scraper): stop setting bogus openclaw/skills githubUrl

[oratis]

Problem

scripts/scrape-skill-details.ts extracted githubUrl from github.com/openclaw/skills/tree/main/skills/<author>/<name> links on clawskills.sh pages — a repo that does not exist (HTTP 404). This had polluted 5028 of 5385 Skill.githubUrl rows in prod. The /api/cron/check-github route was accurately reporting them as dead links.

Root cause

clawskills.sh exposes no per-skill source repo. Verified against the live site and both data-origin scrapers (parse-skills.ts, scrape-clawskills.ts):

• The canonical per-skill link is the ClawHub page (already stored as clawHubUrl / clawSkillsUrl).
• The only GitHub link on a skill page is the site-wide "Star" button to the aggregate list github.com/VoltAgent/awesome-openclaw-skills — not a per-skill source.

So bogus values could not be corrected to real repos (none exist per-skill); they could only be nulled.

Change

The scraper no longer populates githubUrl for clawskills-sourced skills — it's left NULL. Genuine githubUrl values continue to come only from source=GITHUB_SCRAPE skills (scrape-github-skills.ts, real repo.htmlUrl). Removed the interface field, the DOM query, the return field, and the update assignment, with an explanatory comment.

Prod data remediation (already applied, out of band — not in this PR)

A one-time DB update NULLed githubUrl + cleared stale ghStatus/ghCheckedAt on the 5028 affected rows (backup taken; exactly 5028 rows asserted before mutating). Post-state: 0 bogus, 222 real URLs preserved. All affected skills retain their clawSkillsUrl fallback, which the UI (SkillCard) already prefers when githubUrl is null — so no user-visible link was lost.

🤖 Generated with Claude Code

[oratis]

Review — LGTM ✅

改动小而准,root cause 分析扎实:

• 确认了 clawskills.sh 页面上唯一的 GitHub 链接是站级 Star 按钮(聚合列表),github.com/openclaw/skills/... 这个 per-skill 路径根本不存在(404),所以"修正为真实 repo"不可行、只能不写 —— 结论成立。
• 删除路径完整:interface 字段、DOM 提取、返回值、updateData.githubUrl 赋值四处全清,没有残留。留下的注释把"为什么这里不抓 githubUrl"和 5028 行污染的历史讲清楚了,能防止未来有人"好心"把提取逻辑加回来。
• 真实 githubUrl 的唯一来源收敛到 scrape-github-skills.ts(source=GITHUB_SCRAPE),与 prod 修复后的状态(222 条真实 URL 保留)一致。
• UI 侧 SkillCard 在 githubUrl 为 null 时回退 clawSkillsUrl,用户无感。

一个小提醒(不阻塞):prod 数据已在带外 NULL 掉,但如果哪天用旧备份恢复数据库,这批脏数据会回来——备份恢复 runbook 里值得记一笔。

可以合并。