← 返回 Skills 市场
zhbgyj

Knowledge

作者 zhbgyj · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
1178
总下载
0
收藏
20
当前安装
1
版本数
在 OpenClaw 中安装
/install knowledge
功能描述
本地知识库集成 - 文档检索、投喂、双轨模式切换
使用说明 (SKILL.md)

本地知识库

使用方法

命令 说明
"查一下 xxx" 搜索知识库
"切换到本地知识库" 使用本地检索
"切换到 AnythingLLM" 使用对话模式
"知识库统计" 查看文档数量

文件位置

  • 知识库根目录:E:/knowledge-base
  • API服务:http://127.0.0.1:8001
安全使用建议
This skill appears to implement a local knowledge‑base frontend, which is reasonable. Before installing or enabling it, check the following: 1) The skill's openclaw_skill.py imports handle_request from scripts/openclaw_integration that is not included—find and review that module's source (E:/knowledge-base/scripts/openclaw_integration.py) to ensure it does not execute unwanted actions or exfiltrate data. 2) Confirm you trust the local API at http://127.0.0.1:8001 and the configured KB root (E:/knowledge-base); those endpoints/files could be read or written by the skill. 3) Because the skill executes local Python code, consider running it in an isolated environment (VM/container) or obtaining the complete source from a trusted origin. If you cannot inspect the missing module or verify the local service, treat the skill as potentially unsafe.
功能分析
Type: OpenClaw Skill Name: knowledge Version: 1.0.0 The `openclaw_skill.py` file directly passes user-controlled input (extracted as `query`) to the `handle_request` function without explicit sanitization. This `handle_request` function is expected to interact with a local API service (`http://127.0.0.1:8001` as indicated in `SKILL.md`). This pattern creates a significant risk of injection vulnerabilities (e.g., shell injection, API parameter injection) if the `handle_request` implementation or the local API does not properly sanitize or escape the `query` before processing, potentially leading to unauthorized command execution or data manipulation on the local system.
能力评估
Purpose & Capability
Name/description, SKILL.md (E:/knowledge-base root and localhost API) and the code all point to a local KB integration; this is coherent. However the runtime code depends on an external local module scripts/openclaw_integration (imported via sys.path modification) which is not bundled or documented in SKILL.md. The hard-coded Windows path (E:/knowledge-base) is also a platform assumption.
Instruction Scope
SKILL.md instructs use of a local API and KB path and does not disclose that the skill's Python file will import and execute a local module. That imported module (handle_request) is responsible for network/file operations and is absent from the package—its behavior is unknown and could access arbitrary local files or external endpoints.
Install Mechanism
No install spec and no external downloads; the skill is instruction+code only, so nothing extra is pulled from the network by the skill bundle itself.
Credentials
The skill declares no required environment variables or credentials, which is proportionate to a local KB integration. Note: it relies on local filesystem paths and a localhost API endpoint, which are reasonable but sensitive resources.
Persistence & Privilege
The skill does not request always:true or elevated platform privileges and does not modify other skills. The main risk is that it will import/execute code from a local scripts directory which may persist or behave beyond the skill's apparent scope.
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 knowledge
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /knowledge 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
v1.0.0
- 初始版本发布,集成本地知识库功能 - 支持文档检索、知识投喂、双轨(本地/AnythingLLM)模式切换 - 提供常用命令,包括切换模式与知识库统计 - 明确知识库文件位置和API服务地址
元数据
Slug knowledge
版本 1.0.0
许可证
累计安装 22
当前安装数 20
历史版本数 1
常见问题

Knowledge 是什么?

本地知识库集成 - 文档检索、投喂、双轨模式切换. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 1178 次。

如何安装 Knowledge?

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

Knowledge 是免费的吗?

是的,Knowledge 完全免费(开源免费),可自由下载、安装和使用。

Knowledge 支持哪些平台?

Knowledge 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 Knowledge?

由 zhbgyj(@zhbgyj)开发并维护,当前版本 v1.0.0。

💬 留言讨论