Minimax Use

MiniMax AI tools for web search, image analysis, LLM chat, and language translation. Use when you need to search the web, analyze images, generate text with...

Security Scan

The skill is internally consistent with its stated purpose (web search, image analysis, LLM chat, translation) and only requests a single service API key; nothing in the code or instructions appears disproportionate or covert.

Purpose & Capability

Name/description, SKILL.md, references/API.md, assets/models.json, and the Python code all consistently implement web search, image understanding, chat, streaming chat, and translation via a MiniMax HTTP API. The single required environment variable (MINIMAX_API_KEY) and optional MINIMAX_API_HOST are appropriate for an API client. (Minor note: some headers/path naming (e.g. 'anthropic' and 'anthropic-version') suggest the code pattern was adapted from another provider, but this is a code-style artifact, not incoherence.)

Instruction Scope

Runtime instructions and CLI usage match the functions implemented in code. The only file reads the skill performs are explicit (reading a local image file when the user supplies a path) and those contents are intentionally base64-encoded and sent to the configured remote API for image analysis. There are no instructions to scan arbitrary system files, environment variables beyond MINIMAX_API_KEY/MINIMAX_API_HOST, or to exfiltrate unrelated data.

Install Mechanism

No install spec is provided (instruction-only style), and included code is pure Python relying on the requests library. Nothing is downloaded from an untrusted URL and no extract/install steps are present.

Credentials

Only MINIMAX_API_KEY (primary credential) is required, with an optional MINIMAX_API_HOST override. This is proportionate for a remote API client. No unrelated secrets or system config paths are requested.

Persistence & Privilege

The skill is not marked always:true, does not request system-wide configuration changes, and has no elevated persistence. It will only operate when invoked and uses the environment variables the user provides.

Assessment

This skill appears to do what it says: it sends queries, chat messages, and (if you provide a local path) base64-encoded image data to the configured MiniMax API. Before installing: 1) Only provide MINIMAX_API_KEY if you trust the MiniMax provider and the package source; the registry metadata has no homepage, so verify the publisher. 2) Avoid passing sensitive local images or files to the skill (they are uploaded to the remote API). 3) Consider creating an API key with limited scope and rate limits. 4) If you need stronger assurance, inspect the full scripts/__main__.py in your environment and/or run in an isolated environment before granting access to sensitive data or keys.

MiniMax Tools

This skill provides access to MiniMax's AI capabilities, including web search, image analysis, LLM conversations, and text translation.

Setup

First, set up your API key:

export MINIMAX_API_KEY="your-api-key"

To get an API key, sign up at https://platform.minimaxi.com/subscribe/coding-plan

Optionally, you can customize the API endpoint:

export MINIMAX_API_HOST="https://api.minimaxi.com/anthropic"

CLI Commands

# Search the web
python -m scripts web_search "your search query"

# Analyze an image
python -m scripts understand_image "what do you see?" /path/to/image.jpg

# Chat with an LLM
python -m scripts chat "hello, how are you?"

# Stream chat (receive response in chunks)
python -m scripts stream_chat "tell me a story"

# Translate text
python -m scripts translate "hello world" --to Chinese

# List available models
python -m scripts models

Commands Overview

Command	What it does
web_search	Search the web using MiniMax's search API
understand_image	Analyze images using MiniMax's vision model
chat	Have a conversation with MiniMax LLMs
stream_chat	Stream chat with real-time response chunks
translate	Translate text between languages
models	Show all available MiniMax models

CLI Examples

# Search the web
python -m scripts web_search "your search query"

# Analyze an image
python -m scripts understand_image "what do you see?" /path/to/image.jpg

# Chat with an LLM
python -m scripts chat "hello, how are you?"

# Stream chat (receive response in chunks)
python -m scripts stream_chat "tell me a story"

# Translate text
python -m scripts translate "hello world" --to Chinese

# List available models
python -m scripts models

Using in Python

Web Search

from scripts import web_search

result = web_search("Python best practices", count=10)
print(result)

Image Analysis

from scripts import understand_image

# From a local file
result = understand_image("What's in this image?", "/path/to/image.png")

# From a URL
result = understand_image("Describe this image", "https://example.com/image.jpg")

LLM Chat

from scripts import chat

result = chat(
    message="Hello, introduce yourself",
    system="You are a helpful AI assistant",
    model="MiniMax-M2.7",
    temperature=1.0,
    max_tokens=4096,
    stream=False,
    history=[
        {"role": "user", "content": "Hi"},
        {"role": "assistant", "content": "Hello! How can I help you?"}
    ]
)

Parameters:

• message (str): User message
• system (str, optional): System prompt
• model (str, default: "MiniMax-M2.7"): Model name
• temperature (float, default: 1.0): Temperature parameter, range (0.0, 1.0]
• max_tokens (int, default: 4096): Max tokens to generate
• stream (bool, default: False): Enable streaming response
• history (list, optional): History list for multi-turn conversation, each message {"role": "user"/"assistant", "content": "..."}

Streaming Chat

from scripts import stream_chat

result = stream_chat(
    message="Tell me a short story",
    system="You are a storyteller",
    model="MiniMax-M2.7",
    temperature=1.0,
    max_tokens=500
)

# Access streaming chunks
if result["success"]:
    chunks = result["result"]["chunks"]
    full_content = result["result"]["content"]
    print(f"Total chunks: {len(chunks)}")
    print(f"Full content: {full_content}")

Translation

from scripts import translate

result = translate(
    text="Hello World",
    target_lang="Chinese",
    source_lang="auto",
    model="MiniMax-M2.7",
    temperature=1.0,
    max_tokens=4096
)

Parameters:

• text (str): Text to translate
• target_lang (str, default: "English"): Target language, e.g., "English", "Chinese", "Japanese"
• source_lang (str, default: "auto"): Source language, "auto" for auto-detect
• model (str, default: "MiniMax-M2.7"): Model name
• temperature (float, default: 1.0): Temperature parameter, range (0.0, 1.0]
• max_tokens (int, default: 4096): Max tokens to generate

Response Format

All functions return a consistent response format:

Success:

{
  "success": true,
  "result": {...}
}

Error:

{
  "success": false,
  "error": "error message"
}

Learn More

• Full API documentation: references/API.md
• Available models: assets/models.json