← Back to Skills Marketplace
8050
Downloads
6
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install minimax-understand-image
Description
使用 MiniMax MCP 进行图像理解和分析。触发条件：(1) 用户要求分析图片、理解图像、描述图片内容 (2) 需要识别图片中的物体、文字、场景 (3) 使用 MiniMax 的 understand_image 功能
Usage Guidance
Install only after reviewing the setup commands and trusting the MiniMax MCP package source. Prefer installing uv through a trusted package manager, pinning the MCP package version, avoiding mirror-based installs unless approved, and limiting what environment variables are passed to the subprocess. Treat submitted images and prompts as data sent to a third-party service, and avoid using this skill on sensitive screenshots, documents, or photos unless that is acceptable.
Capability Analysis
Type: OpenClaw Skill
Name: minimax-understand-image
Version: 1.0.3
The skill is classified as suspicious primarily due to the `SKILL.md` instructions. It directs the agent to execute `curl -LsSf https://astral.sh/uv/install.sh | sh` for dependency installation, which is a significant supply chain vulnerability as it executes an arbitrary script from the internet without prior inspection. Additionally, `SKILL.md` instructs the agent to read `~/.openclaw/agents/main/agent/auth-profiles.json` to find API keys, which, while intended for a legitimate purpose, could expose sensitive data if the agent is maliciously prompted to misuse this capability. The `understand_image.py` script itself appears to handle inputs and API keys securely via JSON-RPC, but the installation and configuration steps in `SKILL.md` introduce notable risks.
Capability Assessment
Credentials
scripts/understand_image.py starts a third-party MCP subprocess and passes the full parent environment plus the MiniMax API key, which is broader access than needed for image analysis.
Install Mechanism
The skill recommends curl-to-shell installation for uv and installs or invokes minimax-coding-plan-mcp through uvx without a pinned version or checksum, creating mutable supply-chain execution risk.
Instruction Scope
The instructions broadly trigger on image analysis requests and include setup, credential discovery, credential storage, dependency installation, and remote service invocation without clear permission boundaries or a clear privacy warning before images and prompts are sent externally.
Persistence & Privilege
The skill stores the API key persistently in ~/.openclaw/config/minimax.json and looks for MiniMax credentials in an auth profile file. It does not show backdoor persistence or privilege escalation, but credential handling is sensitive and under-scoped.
Purpose & Capability
The stated purpose is MiniMax image understanding, and the script does call a MiniMax MCP image-analysis tool with an image source and prompt. External API use, local image paths or URLs, and a MiniMax API key are coherent with that purpose.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install minimax-understand-image - After installation, invoke the skill by name or use
/minimax-understand-image - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.3
minimax-understand-image 1.0.3
- 流程优化：首次配置流程更为简洁，后续可直接调用图像分析功能。
- API Key 自动获取逻辑修改：默认从 ~/.openclaw/agents/main/agent/auth-profiles.json 查找，而非此前的路径。
- 清理文档内容，去除冗余完整流程示例，提升可读性。
- 步骤描述更明确，帮助用户更快完成配置和调用。
v1.0.2
- 补充了如未购买 MiniMax 时的购买链接指引。
- 其余功能与说明保持不变。
v1.0.1
- Initial skill metadata file (_meta.json) added.
- SKILL.md updated with clearer setup steps and improved API Key acquisition logic.
- Priority for API Key: now tries the environment variable (MINIMAX_API_KEY) before falling back to config file.
- Paths and instructions generalized to use user home directories (e.g., ~/) and current skill directory ({curDir}), improving portability.
- Steps for API Key retrieval now reference newer config file locations and remove reliance on gateway provider extraction.
- Documentation streamlined for easier first-time setup and maintenance.
v1.0.0
Initial release of minimax-understand-image.
- Enables image understanding and analysis using MiniMax MCP.
- Provides clear installation and configuration steps for dependencies and MiniMax API key.
- Supports image input via local path or URL and detailed prompts for analysis.
- Includes example command-line usage and full workflow scenarios.
- Error handling for missing dependencies and API key is documented.
- Script automates setup and invocation of MiniMax's understand_image functionality.
Metadata
Frequently Asked Questions
What is minimax-understand-image?
使用 MiniMax MCP 进行图像理解和分析。触发条件：(1) 用户要求分析图片、理解图像、描述图片内容 (2) 需要识别图片中的物体、文字、场景 (3) 使用 MiniMax 的 understand_image 功能. It is an AI Agent Skill for Claude Code / OpenClaw, with 8050 downloads so far.
How do I install minimax-understand-image?
Run "/install minimax-understand-image" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is minimax-understand-image free?
Yes, minimax-understand-image is completely free (open-source). You can download, install and use it at no cost.
Which platforms does minimax-understand-image support?
minimax-understand-image is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created minimax-understand-image?
It is built and maintained by 要啥自行车 (@thincher); the current version is v1.0.3.
More Skills