Agent Skills › zapprosite/JARVIS-REFRIMIX-ENTERPRISE › hvac_rag_answerer

hvac_rag_answerer

GitHub

基于检索到的HVAC文档片段生成技术回答，强制引用来源并输出结构化JSON。

.agent/skills/hvac_rag_answerer/SKILL.md zapprosite/JARVIS-REFRIMIX-ENTERPRISE

触发场景

需要基于特定文档片段回答技术问题要求答案必须包含引用来源和页码

安装

npx skills add zapprosite/JARVIS-REFRIMIX-ENTERPRISE --skill hvac_rag_answerer -g -y

更多选项

非标准路径

npx skills add https://github.com/zapprosite/JARVIS-REFRIMIX-ENTERPRISE/tree/main/.agent/skills/hvac_rag_answerer -g -y

不安装直接使用

npx skills use zapprosite/JARVIS-REFRIMIX-ENTERPRISE@hvac_rag_answerer

指定 Agent (Claude Code)

npx skills add zapprosite/JARVIS-REFRIMIX-ENTERPRISE --skill hvac_rag_answerer -a claude-code -g -y

安装 repo 全部 skill

npx skills add zapprosite/JARVIS-REFRIMIX-ENTERPRISE --all -g -y

预览 repo 内 skill

npx skills add zapprosite/JARVIS-REFRIMIX-ENTERPRISE --list

SKILL.md

Frontmatter

{
    "name": "hvac_rag_answerer",
    "description": "Gera respostas técnicas baseadas em chunks recuperados com citações obrigatórias"
}

HVAC RAG Answerer Skill

Crie uma função/utilitário que:

Receba: query (str) e retrieved_chunks (list[dict]).

Contexto: Formate os chunks em um prompt:

Use ONLY the following context to answer. If not found, say "I don't know".

Context:
[1] (File: manual_daikin.pdf, Page: 32) ...text...
[2] (File: manual_lg.pdf, Page: 10) ...text...

Prompt System:
- Enforce JSON output: { answer: string, citations: [{ source, page, excerpt }] }.
- Regra: Só inclua citação se usada na resposta.
Output: Retorne o objeto parseado.