AB Agents Vision (MiniMax)

👁️ Image analysis via MiniMax VL API. Describe images, extract text from screenshots, analyze photos. Requires MiniMax Token Plan API key (free tier availab...

Security Scan

The skill mostly does what it says (calls MiniMax to analyze images) but has inconsistencies and a few risky behaviours you should review before installing.

Purpose & Capability

The skill's stated purpose (image analysis via MiniMax) matches the code's behavior (it calls a MiniMax plan via uvx). However the registry metadata declares no required environment variables while SKILL.md and the script require MINIMAX_API_KEY and the uvx binary. That mismatch between declared requirements and actual requirements is an incoherence to be aware of.

Instruction Scope

SKILL.md instructs installation of uvx and setting MINIMAX_API_KEY. The shipped script (vision.sh) reads MINIMAX_API_KEY from the environment but also silently falls back to reading /root/.openclaw/.minimax-env if the env var is not set. That fallback means the skill will try to read an absolute system path (potentially containing other secrets) without declaring it, which expands its data access beyond the described usage.

Install Mechanism

There is no formal install spec in the registry (instruction-only), but SKILL.md recommends installing uvx via a curl | sh from https://astral.sh. Piping remote install scripts to sh is higher risk than using a vetted package manager or an audited release; the install source is not a canonical OS package host and is worth manual review.

Credentials

Requesting a MiniMax API key is appropriate for this purpose. But the registry failing to declare MINIMAX_API_KEY (and the script's secret-file fallback) are disproportionate and unclear. The script's attempt to read /root/.openclaw/.minimax-env could expose other tokens stored there. The skill does not request unrelated cloud credentials, but its undeclared file access is problematic.

Persistence & Privilege

The skill does not request always: true and does not attempt to modify other skills or system-wide settings. Still, the script's reading of an absolute path (/root/.openclaw/.minimax-env) is a form of data access privilege that should be considered before installing or running.

What to consider before installing

Before installing or running this skill: 1) Expect to provide a MINIMAX_API_KEY — the registry metadata omitted this but both SKILL.md and the script require it. 2) Inspect vision.sh yourself (it's included) and decide whether you are comfortable with its fallback that reads /root/.openclaw/.minimax-env; if you don't want that, remove or edit that line so the script only uses the explicit environment variable. 3) Avoid blindly running the recommended curl | sh installer for uvx; instead obtain uvx from a vetted source or review the installer script before executing. 4) Run the skill in a restricted/sandboxed environment if you are unsure, and ensure the MINIMAX_API_KEY you supply is scoped only to the MiniMax service (don't reuse broad or privileged tokens). 5) Prefer that the skill metadata be updated to declare MINIMAX_API_KEY and any other required binaries so the requirements match the actual behavior.

Image analysis via MiniMax VL API — simple, fast, reliable.

⚠️ Requires MiniMax Token Plan API key — get free key

What It Does

• 📸 Describe images — Get detailed scene descriptions
• 📝 Extract text — Read from screenshots, photos, documents
• 🔍 Analyze photos — Identify objects, people, settings
• 🌐 URL support — Analyze images from the web

Requirements

• MiniMax Token Plan API key — Subscribe free
• Linux/macOS
• uvx (auto-installed)

Quick Start

# 1. Install uvx
curl -LsSf https://astral.sh/uv/install.sh | sh

# 2. Get free MiniMax API key
# https://platform.minimax.io → Subscribe → Token Plan (free tier)

# 3. Use
export MINIMAX_API_KEY="sk-cp-your-key"
./vision.sh image.jpg "Describe this image"

Usage

# Basic description
./vision.sh photo.jpg

# With custom prompt
./vision.sh screenshot.png "What text do you see?"

# URL support
./vision.sh "https://example.com/image.jpg" "Describe this"

Examples

Screenshot analysis:

Input: screenshot.png + "What text is in the image?"
Output: "The screenshot shows a code editor with Python code..."

Photo description:

Input: photo.jpg + "Describe in detail"
Output: "A person's bare foot and lower leg resting on a brown
textured waffle-weave blanket. The skin is light-toned..."

Installation

git clone https://github.com/alexburrstudio/ab-agents-vision.git
cd ab-agents-vision/skills/vision
chmod +x vision.sh

Or via ClaWHub:

clawhub install AB-Agents-Vision-MiniMax

Troubleshooting

Error	Solution
API Error: 1033	Retry — MiniMax system error
No response	Check MINIMAX_API_KEY is set correctly
Slow	Use smaller images (<10MB)

---

AB-Agents 🦀

Related Skills

📊 AB Agents Meter Reader — Read meter readings from photos (uses this skill for vision)

---

AB-Agents 🦀