← 返回 Skills 市场
autoxj

DeepWiki Ask

作者 do_while_true · GitHub ↗ · v1.1.0 · MIT-0
cross-platform ⚠ suspicious
355
总下载
1
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install deepwiki-ask
功能描述
通过 DeepWiki MCP 查询仓库信息。支持提问、获取结构、获取文档内容。Query a repository via DeepWiki MCP: ask questions, get structure, get documentation. 用户提供 owner/repo 时触发。
使用说明 (SKILL.md)

DeepWiki 仓库查询

通过 DeepWiki MCP 对指定仓库发起查询,支持三种操作模式:提问、获取仓库结构、获取文档内容。

触发场景

  • 用户询问某仓库的作用、结构或功能
  • 用户提供仓库名(owner/repo)并带有问题
  • 用户需要了解仓库的整体结构
  • 用户需要查看仓库的详细文档内容

参数

参数 必填 说明
repo 仓库名 owner/repo
question 要问的问题(提问模式)
structure 获取文档结构(结构模式)
contents 获取文档内容(内容模式)
topic 指定文档主题(与 contents 一起使用)

执行流程

提问模式

  1. 从用户消息提取 repo(owner/repo)和 question
  2. 执行(必须加 --json):
    python \x3CSKILL_ROOT>/deepwiki_ask.py -r \x3Cowner/repo> -q "\x3Cquestion>" --json
    
    Windows 下中文问题若编码异常,可把问题写入 UTF-8 文件后:-q @\x3CSKILL_ROOT>/temp_q.txt

结构模式

  1. 从用户消息提取 repo(owner/repo)。
  2. 执行(必须加 --json):
    python \x3CSKILL_ROOT>/deepwiki_ask.py -r \x3Cowner/repo> --structure --json
    

内容模式

  1. 从用户消息提取 repo(owner/repo)和可选的 topic

  2. 执行(必须加 --json):

    python \x3CSKILL_ROOT>/deepwiki_ask.py -r \x3Cowner/repo> --contents --json
    python \x3CSKILL_ROOT>/deepwiki_ask.py -r \x3Cowner/repo> --contents --topic "\x3Ctopic_name>" --json
    
  3. 解析 stdout JSON:status == "success" 则根据操作模式展示相应结果;status == "error" 则提示 message

  4. 请求可能需 30–120 秒,需等待。

输出示例

提问模式

{"status": "success", "repo": "owner/repo", "mode": "question", "question": "...", "result": "..."}

结构模式

{"status": "success", "repo": "owner/repo", "mode": "structure", "result": "..."}

内容模式

{"status": "success", "repo": "owner/repo", "mode": "contents", "result": "..."}

错误响应

{"status": "error", "repo": "owner/repo", "message": "..."}

配置

config.jsonrequest_timeout_seconds(10–600,默认 120)、request_max_retries(0–10,默认 3)。

错误处理

  • 仓库格式错误:提示 owner/repo 格式
  • 超时/网络错误:脚本重试后返回 status: "error",需要提示用户检查网络

历史版本

v1.1.0 (2026-03-14)

  • 📋 支持获取文档结构(--structure)
  • 📄 支持获取文档内容(--contents)
  • 📄 支持指定文档主题(--topic)
  • 🔄 重构为 MCP 客户端类,移除 requests 依赖,改用标准库 urllib

v1.0.0 (2026-03-10)

  • 🎉 初始版本发布
  • 📖 支持 owner/repo + question 单次问答
安全使用建议
This skill appears to do what it says: it runs a local Python client that sends the user-supplied owner/repo and question to DeepWiki's MCP endpoint (https://mcp.deepwiki.com/mcp) and returns results. Before installing, consider: 1) Privacy — repository names and any question text will be transmitted to an external service; do not submit secrets or private data. 2) Trust the endpoint — verify the DeepWiki domain and organization if you need to trust data handling. 3) Local file writes — the script may create or update config.json in the skill folder. 4) Long network calls — requests can take up to 30–120s per SKILL.md. 5) Incomplete source provided here — the file contents in the review were truncated, so review the full deepwiki_ask.py source locally to confirm there are no additional network endpoints, credential harvesting, telemetry, or unexpected behavior before enabling autonomous invocation. If you want higher assurance, run the script in a sandbox and inspect outbound network traffic.
功能分析
Type: OpenClaw Skill Name: deepwiki-ask Version: 1.1.0 The skill bundle provides a legitimate interface for querying repository documentation via the DeepWiki MCP service. The Python script (deepwiki_ask.py) implements a standard MCP client using only the Python standard library (urllib) to communicate with https://mcp.deepwiki.com/mcp, and the instructions in SKILL.md are clearly aligned with the stated purpose of repository analysis without any signs of malicious intent or prompt injection.
能力评估
Purpose & Capability
Name/description describe querying DeepWiki MCP for repository info; the code implements an MCP JSON-RPC client that POSTs queries to https://mcp.deepwiki.com/mcp and exposes methods for question/structure/contents. No unrelated binaries, env vars, or services are requested.
Instruction Scope
SKILL.md instructs the agent to run the included Python script with --json and parse stdout. That matches the script's intended CLI usage. The runtime behavior includes network calls to an external DeepWiki MCP endpoint using the repo and question provided by the user — expected for this skill, but it does transmit user inputs outside the host.
Install Mechanism
No install spec; this is instruction + script. No downloads, package installs, or archive extraction are present in the manifest.
Credentials
The skill requests no environment variables or credentials. It reads/writes a local config.json in its own directory (creates default if missing). All accessed configuration is proportional to the described functionality.
Persistence & Privilege
always is false, the skill does not request elevated or cross-skill configuration. It may create a local config.json (expected) but does not modify other skills or system-wide settings according to the visible files.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install deepwiki-ask
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /deepwiki-ask 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.1.0
Expanded to support repository structure and documentation queries: - Added structure mode: retrieve repository documentation structure with --structure. - Added contents mode: retrieve detailed documentation content with --contents, supports specifying a topic. - Updated parameters and output formats to handle multiple query modes. - Refined description to clarify function and use scenarios. - Maintained previous Q&A functionality; improved error handling instructions.
v1.0.0
Initial release of deepwiki-ask - Query any GitHub repository with a single question using DeepWiki MCP. - Triggered when a user provides an owner/repo and a question. - Does not store question history; each request is independent. - Supports error handling for invalid repo format and connection issues. - Customizable timeout and retry settings via config.json. - Response includes either an answer or an error message.
元数据
Slug deepwiki-ask
版本 1.1.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

DeepWiki Ask 是什么?

通过 DeepWiki MCP 查询仓库信息。支持提问、获取结构、获取文档内容。Query a repository via DeepWiki MCP: ask questions, get structure, get documentation. 用户提供 owner/repo 时触发。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 355 次。

如何安装 DeepWiki Ask?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install deepwiki-ask」即可一键安装,无需额外配置。

DeepWiki Ask 是免费的吗?

是的,DeepWiki Ask 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

DeepWiki Ask 支持哪些平台?

DeepWiki Ask 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 DeepWiki Ask?

由 do_while_true(@autoxj)开发并维护,当前版本 v1.1.0。

💬 留言讨论