classify-structured

Agent Skills › zerogpu/zerogpu-router › classify-structured

classify-structured

GitHub

基于JSON Schema的多维度文本分类技能，支持同时按多个轴（如情感、主题）对文本进行分类，自动映射允许标签并输出结构化结果。

agents/claude/skills/classify-structured/SKILL.md zerogpu/zerogpu-router

Trigger Scenarios

需要对文本进行多维度分类指定了多个分类轴和标签集合

Install

npx skills add zerogpu/zerogpu-router --skill classify-structured -g -y

More Options

Non-standard path

npx skills add https://github.com/zerogpu/zerogpu-router/tree/main/agents/claude/skills/classify-structured -g -y

Use without installing

npx skills use zerogpu/zerogpu-router@classify-structured

指定 Agent (Claude Code)

npx skills add zerogpu/zerogpu-router --skill classify-structured -a claude-code -g -y

安装 repo 全部 skill

npx skills add zerogpu/zerogpu-router --all -g -y

预览 repo 内 skill

npx skills add zerogpu/zerogpu-router --list

SKILL.md

Frontmatter

{
    "name": "classify-structured",
    "description": "Multi-axis classification using a JSON schema mapping categories to allowed labels (gliner2-base-v1). Use when the user wants to classify text along several dimensions at once, e.g. \"classify by sentiment and topic\" with explicit label sets per axis.",
    "allowed-tools": "Bash(zerogpu chat_completions *)",
    "argument-hint": "<text> -s '<json schema>'"
}

Run schema-driven classification. The request below holds the text and the schema — each axis with its allowed labels — given as -s '<json>' or in plain words.

Run this with the Bash tool, putting the schema in --metadata and pasting the text into the heredoc verbatim — no escaping, the quoted heredoc handles every shell metacharacter, newline, quote, and paren:

zerogpu chat_completions -m gliner2-base-v1 --metadata '{"usecase":"classification","schema":{"sentiment":["positive","negative","neutral"],"topic":["support","billing","product"]}}' <<'ZGPU_END_OF_INPUT'
<the source text, verbatim>
ZGPU_END_OF_INPUT

"usecase":"classification" is required exactly as written; the API rejects the request without it.
schema maps each axis to its allowed labels. If the request gives axes without labels, pick sensible label sets and say which you used.
--metadata is single-quoted JSON, so write any ' inside a label as '\''.

Output is a JSON object whose classification holds one chosen label per axis.

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.

Request: $ARGUMENTS