classify-domain
GitHub根据域名分类其所属业务领域，无需获取页面内容。适用于域名审核、白名单评分及内容归类场景。
Trigger Scenarios
Install
npx skills add zerogpu/zerogpu-router --skill classify-domain -g -y
SKILL.md
Frontmatter
{
"name": "classify-domain",
"description": "Classify a domain name against the IAB taxonomy without fetching the page (zlm-v1-iab-domain-classifier). Use when the user has a hostname or URL rather than article text — bidstream enrichment, allow\/deny-list scoring, \"what is example.com about?\".",
"allowed-tools": "Bash(zerogpu classify_domain*)",
"argument-hint": "<domain>"
}
Classify a domain:
zerogpu classify_domain "$ARGUMENTS"
Format $ARGUMENTS as a bare hostname. The model takes only the domain — strip the scheme, path, query, and fragment first. https://www.nytimes.com/section/world?x=1 becomes www.nytimes.com. If the user gave you several domains, run the command once per domain.
Output is a JSON object with audience (IAB 2.2 audience tiers with scores) and content (iab_1_0 and iab_2_2 category lists).
Use this instead of /zerogpu-router:classify-iab when the input is a domain — it needs only the hostname, so payloads are up to 10x smaller than sending page text. If the user has the actual article text and wants per-URL precision, use classify-iab or classify-iab-enriched.
Savings note: only if the command output literally contains a line starting with 💰 ZeroGPU savings, append that exact line, unchanged, as the last line of your reply. If no such line is present, say nothing about savings and do not mention or suggest /zerogpu-router:cost-savings — this note is intentionally occasional, not shown every time.
Version History
- cee9321 Current 2026-08-02 21:14
