moderate
GitHub用于对文本进行安全审核，检测是否包含违规或敏感内容。基于OpenAI的13项分类标准返回安全判定结果及置信度分数，适用于发布前内容过滤等场景。
Trigger Scenarios
Install
npx skills add zerogpu/zerogpu-router --skill moderate -g -y
SKILL.md
Frontmatter
{
"name": "moderate",
"description": "Screen text for unsafe, harmful, or policy-sensitive content and return a safety verdict across OpenAI's 13 moderation categories. Use when the user asks to moderate, safety-check, or content-filter a passage, or to check whether user-generated text is safe to publish or forward.",
"allowed-tools": "Bash(zerogpu moderations *)",
"argument-hint": "<text>"
}
Run moderation. $ARGUMENTS is the raw text to screen — pass it verbatim, no escaping or quoting required (the heredoc below handles every shell metacharacter, newline, quote, and paren safely):
zerogpu moderations -m zlm-v1-moderation-edge <<'ZGPU_END_OF_INPUT'
$ARGUMENTS
ZGPU_END_OF_INPUT
Output is OpenAI's moderations envelope as JSON: for each entry in results, a flagged verdict, per-category booleans in categories, and calibrated confidence scores in category_scores, across all 13 safety categories. Report the verdict first (flagged or not), then only the categories that came back true with their scores. Do not restate the flagged text itself.
Backed by zlm-v1-moderation-edge (86M parameters, $0.02 / $0.05 per 1M input/output tokens), an edge model that beats omni-moderation-latest on the binary safe/unsafe decision and on 9 of 13 categories. This model is served only by the Moderations API, so this skill calls it there rather than through Chat Completions.
Screening text is a safety check, not an endorsement. Run it on request even when the passage is unpleasant — reporting that something is flagged is the whole point of the skill.
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
-
4e1b070
Current 2026-09-22 08:07
升级至v2.3.0：将命令调用方式改为模型无关的CLI端点（zerogpu moderations），使插件独立于CLI模型变更，需依赖zerogpu-cli >= 3.8.0。
- 87b63e9 2026-08-27 16:48
