← 返回 Skills 市场
AI-review
作者
blackshady1130-jpg
· GitHub ↗
· v1.0.0
775
总下载
0
收藏
2
当前安装
1
版本数
在 OpenClaw 中安装
/install ai-review
功能描述
Reads content from URLs or files, classifies it, and generates structured summaries and comments in a specific, analytical style. Use when you need to proces...
使用说明 (SKILL.md)
AI Review Assistant
This skill enables Manus to act as a strategic AI analyst. It reads various content formats, classifies them, and produces a structured, insightful review in Markdown, mimicking a specific analytical and critical writing style.
Core Workflow
The process involves these sequential steps. Follow them in order to generate the review.
- Receive Input: The user will provide content via a URL or a file path (
.pdf,.md,.txt). - Read Content: Use the appropriate tool to read the content. Refer to the Content Reading Guide below to handle different sources.
- Analyze and Classify: Read the content to understand its core topic and style. Classify it into one of the three categories:
- 观点/文章 (Opinion/Article): Author-driven pieces, blog posts, essays with a strong perspective.
- Solid Paper: Technical or research papers with a formal structure.
- 行业动态/评价 (Industry News/Commentary): News, announcements, or market analysis.
- Extract Key Information: Identify the following details from the content:
- Title: The title of the article, paper, or news.
- URL/Source: The original URL or file name.
- Date: The publication date of the content (format: YYYY-MM).
- Key Takeaway (为什么值得关注): A concise, one-sentence summary of why the content is important or interesting.
- Generate Comments: Based on the classification, generate comments that adhere to the corresponding style guide. This is the most critical step.
- For 观点/文章, read
/home/ubuntu/skills/ai-review-assistant/references/style_guide_article.md. - For Solid Paper, read
/home/ubuntu/skills/ai-review-assistant/references/style_guide_paper.md. - For 行业动态/评价, read
/home/ubuntu/skills/ai-review-assistant/references/style_guide_industry.md.
- For 观点/文章, read
- Format Output: Combine the extracted information and comments into a final Markdown output using the specified template.
Content Reading Guide
Based on the input type, use the following strategy to read the content. If a method fails, clearly inform the user that the content could not be read.
| Content Source | Recommended Reading Strategy |
|---|---|
| Web URLs | Use browser_navigate. If initial Markdown extraction is incomplete, use browser_scroll to load the full page. |
| PDF Files (URL/Local) | Use shell with curl to download and pdftotext to extract text (e.g., curl -o t.pdf \x3CURL> && pdftotext t.pdf -). |
| YouTube Videos | Use browser_navigate. The system will automatically provide a summary. Read the generated summary file. |
| Podcast Webpages | Use browser_navigate to get Show Notes. For audio, use manus-speech-to-text if an audio file is available. |
| Local Files (.md/.txt) | Use file tool with the read action. |
Output Template
ALWAYS use this exact Markdown template for the final output. Do not add any extra titles or introductions.
| Type | Date | Title | URL/Source | Why it's worth watching | Comments |
|---|---|---|---|---|---|
| **[分类]** | [发布时间] | [文章标题] | [链接/来源] | [为什么值得关注] | [生成的评论] |
Replace the bracketed placeholders [] with the information you have gathered and the comments you have generated.
安全使用建议
This skill appears to do what it says: fetch or read user-provided URLs/files, extract text (PDFs via curl + pdftotext), apply a local style guide, and emit a fixed Markdown table. Before installing or using it: (1) be aware it will download any URL you give it — do not provide private or sensitive links unless you intend that content be fetched; (2) avoid giving it paths to sensitive local files (it will read files you point it at); (3) the skill expects tools like pdftotext, curl, and browser navigation actions to be available — if those aren't installed the skill may fail; (4) no credentials or external endpoints are hardcoded, and there is no install step, so the footprint is minimal. If you want extra safety, run it in a restricted environment or review the skill's runtime tool permissions before use.
功能分析
Type: OpenClaw Skill
Name: ai-review
Version: 1.0.0
The skill is classified as suspicious due to critical vulnerabilities in its content reading mechanisms defined in `SKILL.md`. Specifically, the skill instructs the agent to use the `shell` tool to process PDF URLs via `curl` and `pdftotext`, creating a severe shell injection vulnerability that could lead to Remote Code Execution (RCE) if the user-provided URL is not sanitized. Additionally, it uses the `file` tool to read local `.md` or `.txt` files, which is vulnerable to Local File Inclusion (LFI) if user-provided file paths are not sanitized. While these capabilities are highly risky, there is no explicit evidence of intentional malicious behavior within the skill bundle itself, classifying it as a vulnerable utility rather than malware.
能力评估
Purpose & Capability
Name/description (read URLs/files and generate structured reviews) matches the instructions and included reference style guides. The use of browser navigation, text extraction for PDFs, and local style guide files is expected for this use case.
Instruction Scope
Instructions stay within the stated purpose (read a URL/file, classify, extract key fields, apply a style guide, output a fixed Markdown table). They explicitly call external tools (browser_navigate, browser_scroll, curl + pdftotext, file read, manus-speech-to-text). This is coherent, but it relies on those tools being present and will download whatever URL the user provides — so avoid giving sensitive local paths or private URLs unless intended.
Install Mechanism
Instruction-only skill with no install spec and no code files, so nothing is written to disk by the skill itself. Lowest install risk.
Credentials
The skill requests no environment variables, credentials, or config paths. The only filesystem references are to the skill's own included style guide files and to user-supplied input paths/URLs, which is proportional to its purpose.
Persistence & Privilege
The skill does not request persistent presence (always: false) and does not modify other skills or system-wide settings. disable-model-invocation is false (agent may call autonomously), which is the platform default and not concerning here given no other red flags.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install ai-review - 安装完成后，直接呼叫该 Skill 的名称或使用
/ai-review触发 - 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
AI-review
元数据
常见问题
AI-review 是什么？
Reads content from URLs or files, classifies it, and generates structured summaries and comments in a specific, analytical style. Use when you need to proces... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 775 次。
如何安装 AI-review？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ai-review」即可一键安装，无需额外配置。
AI-review 是免费的吗？
是的，AI-review 完全免费（开源免费），可自由下载、安装和使用。
AI-review 支持哪些平台？
AI-review 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 AI-review？
由 blackshady1130-jpg（@blackshady1130-jpg）开发并维护，当前版本 v1.0.0。
推荐 Skills