← Back to Skills Marketplace
438
Downloads
0
Stars
4
Active Installs
1
Versions
Install in OpenClaw
/install rag-ingest
Description
将 Agent 已解读好的正文写入 Qdrant（kb_main）。仅做 chunk、embedding 和向量写入，不负责抓取与精炼。
Usage Guidance
此技能在功能和权限上自洽：它需要 QDRANT_URL（你的 Qdrant 实例）和一个嵌入服务的 API key。注意两点：1) 脚本会接受多种环境变量名（VECTORENGINE_API_KEY、EMBED_API_KEY、OPENAI_API_KEY）并有 EMBED_BASE_URL 的默认值（https://api.vectorengine.ai/v1），请确认你信任所用的嵌入服务和对应的 API key；2) SKILL.md 中应明确列出所有可识别的环境变量与默认嵌入端点，以防凭据被意外指向其他服务。建议在受控/隔离环境中先用测试数据和本地 Qdrant 实例（或验证过的远端）运行，确认行为符合预期，然后再在生产环境提供真实 API 密钥。
Capability Analysis
Type: OpenClaw Skill
Name: rag-ingest
Version: 1.0.0
The rag-ingest skill is a utility designed to chunk text, generate embeddings via an external API, and store the resulting vectors in a Qdrant database. The script (scripts/ingest.mjs) follows standard procedures for RAG (Retrieval-Augmented Generation) workflows, using environment variables for configuration and providing clear CLI parameters. No evidence of malicious intent, data exfiltration, or unauthorized execution was found; the network calls to the embedding provider (api.vectorengine.ai) and Qdrant are consistent with the tool's documented purpose.
Capability Assessment
Purpose & Capability
名字与描述表明仅做 chunk、embedding 与写入 Qdrant；脚本正是读取正文（参数或 stdin）、切分、调用嵌入 API、建立/删除/写入 Qdrant collection，所需二进制 node 与声明一致，未请求与用途无关的外部凭据或路径。
Instruction Scope
SKILL.md 与脚本指令一致：通过 CLI/STDIN 接收内容并写入 Qdrant。脚本只访问命令行参数、stdin、环境变量和网络（嵌入服务与 Qdrant），没有读取本地敏感路径或要求额外上下文。网络访问限于嵌入端点（可配置）和 Qdrant_URL。
Install Mechanism
无安装说明（instruction-only + 附带脚本），也没有从不可信 URL 下载或在安装阶段写入磁盘的操作；仅要求运行时有 node，风险较低。
Credentials
声明的必需 env 为 QDRANT_URL 与 EMBED_API_KEY，与用途相符。但脚本还会读取/接受额外环境变量（VECTORENGINE_API_KEY、OPENAI_API_KEY、EMBED_BASE_URL、RAG_INGEST_EMBED_MODEL / OPENAI_EMBEDDING_MODEL）。这些额外名称在 SKILL.md 中未完全列出——功能上可解释（方便兼容多种嵌入提供者），但应在元数据/文档中明确，以避免意外使用未预期的凭据或外部端点（默认 EMBED_BASE_URL 指向 api.vectorengine.ai）。
Persistence & Privilege
技能未声明 always:true，也不修改其他技能或系统配置；默认允许模型调用且为用户可调用，权限与其功能一致。
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install rag-ingest - After installation, invoke the skill by name or use
/rag-ingest - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of rag-ingest.
- Writes pre-processed document content into Qdrant (kb_main collection) with chunking, embedding, and vector storage.
- Does not handle fetching or refinement of content (expects already-processed input).
- Accepts content via --content option or from stdin.
- Supports key parameters: --doc-id, --topic-tags, optional --source, and --collection.
- Designed for integration with tools like deep-research/summarize.
Metadata
Frequently Asked Questions
What is rag-ingest?
将 Agent 已解读好的正文写入 Qdrant（kb_main）。仅做 chunk、embedding 和向量写入，不负责抓取与精炼。 It is an AI Agent Skill for Claude Code / OpenClaw, with 438 downloads so far.
How do I install rag-ingest?
Run "/install rag-ingest" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is rag-ingest free?
Yes, rag-ingest is completely free (open-source). You can download, install and use it at no cost.
Which platforms does rag-ingest support?
rag-ingest is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created rag-ingest?
It is built and maintained by Seal-Re (@seal-re); the current version is v1.0.0.
More Skills