ingest_manual_pdf

GitHub

将技术手册PDF解析为文本块，生成嵌入向量并存储至Qdrant向量数据库，支持去重与元数据管理。

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

需要处理PDF文档入库构建RAG知识库的数据摄取

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

More Options

Non-standard path

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

Use without installing

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

指定 Agent (Claude Code)

npx skills add zapprosite/JARVIS-REFRIMIX-ENTERPRISE --skill ingest_manual_pdf -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

Frontmatter

{
    "name": "ingest_manual_pdf",
    "description": "Pipeline para ingerir PDF de manual técnico"
}

Ingest Manual PDF Skill

Crie script/função que:

Parse: Use pypdf ou unstructured para extrair texto + metadados (page_number).
Chunking: RecursiveCharacterTextSplitter (chunk_size=1000, overlap=200).
Embedding: Gere embeddings (OpenAI ou local).
Upsert Qdrant:
- Collection: manuals
- Payload: { "filename": "...", "page_number": int, "text": "...", "tenant_id": "public" }
- UUID: Gere determinístico (hash do texto) para evitar duplicatas.