MiniMax PDF Analysis V2

Analyze PDF files using MiniMax API. Supports text extraction, keyword search, and image-based VLM analysis (converts PDF pages to images first). Requires Mi...

Security Scan

The skill mostly matches its stated purpose (PDF text/search and image-based VLM analysis) but includes extra undocumented network-capable tooling and will upload local file contents to a remote API — exercise caution before installing or using with sensitive files or a privileged API key.

Purpose & Capability

Name/description (PDF extraction, search, VLM) align with the included Python scripts. Required binaries (python, node) and MINIMAX_API_KEY are reasonable for the VLM feature. However, the bundled Node tool exposes an additional 'web_search' capability and a generic 'understand_image' CLI that can POST arbitrary images or local files to api.minimax.io — this web-search functionality is not documented in SKILL.md and expands the skill's surface beyond the stated PDF-focused purpose. README also references a different domain (platform.minimaxi.com) which is inconsistent with api.minimax.io.

Instruction Scope

SKILL.md instructs running the Python scripts and setting MINIMAX_API_KEY, which is expected. But the codebase contains a Node tool that can perform web searches and convert arbitrary local files to base64 and send them to the MiniMax API. The SKILL.md does not warn that running the Node tool or some modes will upload local file contents (including arbitrary file paths) to a remote service; that could result in inadvertent leakage of sensitive local data if a user or agent invokes those tools.

Install Mechanism

No remote install/downloads or archive extraction are specified; the skill is instruction+source-only. No external installers are executed by the manifest. That lowers supply-chain risk compared with arbitrary downloads.

Credentials

Only MINIMAX_API_KEY is required, which is proportional to calling MiniMax APIs. Be aware that the provided API key will be sent to api.minimax.io when using VLM or web_search; any content you send (images, prompts, extracted text) will be transmitted to that service. The README/SKILL.md request a 'Coding Plan' key (sk-cp-*) which is consistent with the code, but domain references are inconsistent which merits verification.

Persistence & Privilege

The skill does not request always:true, does not alter other skills or system-wide config, and is user-invocable by default. It does not request persistent elevated privileges.

What to consider before installing

This skill appears to implement the advertised PDF extraction/search and image-VLM flows, but it also includes a Node CLI tool that can: (1) perform web searches via the Coding Plan API, and (2) read local image files (or any file path you pass) and upload them as base64 to api.minimax.io. Before installing or using: - Do not provide your primary/privileged API key unless you trust api.minimax.io; consider a restricted or throwaway key for testing. - Avoid running the vision/web_search tools on sensitive PDFs or arbitrary local file paths (they will upload file contents). - Verify the API host and key provisioning URL (manifest references api.minimax.io; README references platform.minimaxi.com — confirm correct endpoints). - If you only need offline text extraction/search, run the Python extract/search scripts without setting the API key or invoking vision. - If you plan to allow autonomous agent invocation, know that an agent could call the Node tool or vision mode and thereby send local file contents to the remote service; restrict agent permissions or review invocation policies. If you want greater assurance, inspect and run the scripts in an isolated environment and audit network calls (e.g., with a firewall) before using with sensitive data.

MiniMax PDF Analysis

使用 MiniMax API 分析 PDF，支援三種模式。

三種分析模式

1. 萃取文字 extract

python scripts/analyze_pdf.py extract input.pdf
python scripts/analyze_pdf.py extract input.pdf --output out.txt
python scripts/analyze_pdf.py extract input.pdf --pages 1-3

2. 搜尋關鍵字 search

python scripts/analyze_pdf.py search input.pdf "PID"
python scripts/analyze_pdf.py search input.pdf "error" --ignore-case
python scripts/analyze_pdf.py search input.pdf "regex.*pattern" --regex

3. VLM 圖片分析 vision

# 需要 MINIMAX_API_KEY 環境變數
export MINIMAX_API_KEY=sk-cp-xxx

# 分析整份 PDF（每頁轉圖片後分析）
python scripts/analyze_pdf.py vision input.pdf "摘要這張投影片的內容"

# 只分析特定頁面
python scripts/analyze_pdf.py vision input.pdf "摘要" --pages 1-3 --dpi 200

環境需求

套件	安裝方式
PyMuPDF	pip install pymupdf
MiniMax API Key	MINIMAX_API_KEY=sk-cp-xxx