Ragtop Agent

高级 RAG 助手，具备 Agentic RAG 思考能力。能够自动化管理 RAGTOP 知识库，并在执行深度调研任务时，调用 RAGTOP 后端接口。使用场景包括：(1) 列出知识库和文档，(2) 执行语义检索和深度分析。

Security Scan

The skill's declared purpose (calling a RAGTOP backend to list KBs/docs and run retrieval) matches its instructions, but there are inconsistencies and a risky default (an internal IP base URL) plus an undeclared environment variable that you should verify before use.

Purpose & Capability

Name and description claim RAGTOP knowledge‑base management and semantic retrieval; the instructions only perform list_kb, list_doc, and retrieval calls to a RAGTOP API — that matches. Minor inconsistency: metadata/registry only declares RAGTOP_API_TOKEN as a required env, but SKILL.md additionally requires RAGTOP_API_URL (with a default).

Instruction Scope

SKILL.md is instruction-only and explicitly tells the agent to run curl against RAGTOP endpoints and to follow a defined multi-step workflow; it does not ask to read unrelated local files or other credentials. However, it defaults the API URL to http://10.71.10.71:9380 (an internal IP) if RAGTOP_API_URL is unset — this means an unstated network target will be contacted unless the user overrides it, which could be risky if the default is not the user's intended endpoint.

Install Mechanism

No install spec and no code files — instruction-only skill. This minimizes disk persistence and arbitrary code installation risk.

Credentials

The skill requests a single credential (RAGTOP_API_TOKEN) which is appropriate for an API‑calling skill. But SKILL.md also uses RAGTOP_API_URL (not declared in registry metadata) and provides a default internal IP — the registry should list RAGTOP_API_URL as an env dependency. Verify the token's scope/privileges before granting it.

Persistence & Privilege

always is false and no config paths or system-wide changes are requested. The skill has no special persistence or elevated privilege.

What to consider before installing

This skill is largely coherent with its stated purpose, but take these precautions before installing or enabling it: - Verify the RAGTOP_API_URL you intend the skill to call; the SKILL.md defaults to an internal IP (http://10.71.10.71:9380). If that is not your RAGTOP service, set RAGTOP_API_URL explicitly — otherwise the agent will make network calls to that address. - Limit the RAGTOP_API_TOKEN privileges (least privilege) and use a token scoped only to the KBs/doc access needed. Prefer short‑lived tokens and rotation. - Ask the publisher for source/homepage or provenance; the registry metadata is missing RAGTOP_API_URL and there is no homepage, which reduces transparency. - Monitor network traffic and logs after enabling, and consider running the agent in a restricted environment until you confirm behavior. If you cannot confirm the default endpoint or token scope, treat this skill as risky and do not supply sensitive credentials.

ragtop-agent Skill

本 Skill 允许 AI 通过 curl 调用 RAGTOP 后端接口。AI 必须根据用户请求的性质，在“简单指令”与“深度调研”两种模式间切换。

Configuration

The following environment variables are required:

• RAGTOP_API_URL: RAGTOP API base URL. Defaults to http://10.71.10.71:9380 if not set.
• RAGTOP_API_TOKEN: Your RAGTOP API access token. Can be configured via the OpenClaw Web UI.

1. 核心工具构建指南 (How to build curl)

在调用以下接口前，请确保已获取环境变量 ${RAGTOP_API_URL} 和 ${RAGTOP_API_TOKEN}。如果 ${RAGTOP_API_URL} 为空，请使用默认值 http://10.71.10.71:9380。

A. 列出知识库 (list_kb)

用于获取所有可用的 knowledge_id。

curl -L -X POST "${RAGTOP_API_URL}/api/v1/ragtop/tool/list_kb" \
     -H "Authorization: Bearer ${RAGTOP_API_TOKEN}" \
     -H "Content-Type: application/json"

B. 列出文档 (list_doc)

用于获取特定知识库内的 doc_id 列表，以便缩小检索范围。

• Payload: {"knowledge_id": "string"}

curl -L -X POST "${RAGTOP_API_URL}/api/v1/ragtop/tool/list_doc" \
     -H "Authorization: Bearer ${RAGTOP_API_TOKEN}" \
     -H "Content-Type: application/json" \
     -d '{"knowledge_id": "YOUR_KB_ID"}'

C. 内容检索 (retrieval)

用于执行语义搜索。支持单查询或多查询。

• Payload 关键参数:
  • knowledge_id: 必填。
  • query: 字符串（单查询）。
  • queries: 字符串数组（多查询，推荐用于复杂任务）。
  • doc_ids: 数组（可选，限定文件范围）。
  • retrieval_setting: {"top_k": 16, "score_threshold": 0.3}

curl -L -X POST "${RAGTOP_API_URL}/api/v1/ragtop/tool/retrieval" \
     -H "Authorization: Bearer ${RAGTOP_API_TOKEN}" \
     -H "Content-Type: application/json" \
     -d '{
       "knowledge_id": "YOUR_KB_ID",
       "queries": ["查询1", "查询2"],
       "retrieval_setting": {"top_k": 5}
     }'

2. 任务分类处理逻辑

情况 A：简单指令 (Simple Instructions)

适用场景: 用户询问“有哪些知识库？”、“这个库里有哪些文件？”等管理类问题。 执行逻辑:

1. 直接根据需求构建 list_kb 或 list_doc 的 curl 命令。
2. 将返回的 JSON 结果整理成易读的表格或列表告知用户。

情况 B：深度调研 (Deep Investigation / Agentic RAG)

适用场景: 用户提出具体业务问题、对比分析或需要跨文档总结。 执行逻辑:

1. 参考 references/workflow.md 执行“分析-分解-检索-综合”流程。
2. 多步编排:
   • 第一步：调用 list_kb 确定相关的知识库 ID。
   • 第二步（可选）：调用 list_doc 锁定相关文件。
   • 第三步：构建包含多个改写问题的 retrieval 请求，利用多路召回提高准确率。
   • 第四步：根据检索到的多个 chunks 进行逻辑推理和引用标注。

3. 运行规范

• 严禁提及 ragflow，统一使用 ragtop。
• 引用必填: 所有深度调研的回答必须注明引用的文档名称。
• 错误处理: 如果 curl 返回非 SUCCESS，应检查 Token 有效性并告知用户。