← 返回 Skills 市场

AB Agents Meter Reader

Name: AB Agents Meter Reader
Author: alexburrstudio

作者 alexburrstudio · GitHub ↗ · v1.0.2 · MIT-0

cross-platform ⚠ suspicious

总下载

当前安装

版本数

在 OpenClaw 中安装

/install ab-agents-meter-reader

功能描述

📊 Read meter readings from photos. Electricity (day/night tariffs) and water meters. Saves history and generates messages for landlord.

使用说明 (SKILL.md)

AB Agents Meter Reader 📊

Read meter readings from photos — electricity and water meters.

Features

⚡ Read electricity meters (single or dual tariff)
💧 Read water meters (hot and cold)
📝 Save readings history with dates
📨 Generate message for landlord
🔄 Track multiple apartments

Setup

Requirements

MiniMax Token Plan API key (for vision)
Linux/macOS

Quick Start

# First run - it will ask questions
./meter-reader.sh

# Later runs - just send photos
./meter-reader.sh photo.jpg

First Run Setup

The script will ask:

Tenant name — your name
Apartment address — full address
Meter layout — how to tell meters apart:
- "left=hot,right=cold" (default)
- Or custom description

Supported Meter Types

Electricity

Single tariff
Dual tariff (day/night) — T1=day, T2=night
Multi-tariff (cycle through screens)

Water

Cold water (usually on left)
Hot water (usually on right)
Cubic meters (m³)

Usage

# Interactive mode (asks for photo)
./meter-reader.sh

# With photo
./meter-reader.sh /path/to/meter.jpg

# Generate message for landlord
./meter-reader.sh --message

How It Works

Analyzes photo using MiniMax VL API
Identifies meter type automatically
Reads the numbers
Saves to readings history
Generates formatted message

History

Readings saved to: ~/.meter-readings/history.json

Format:

{
  "apartments": {
    "address": {
      "tenant": "Name",
      "layout": "left=hot,right=cold",
      "readings": [
        {"date": "2026-04-26", "electricity_day": 8495, "electricity_night": 3008, "water_cold": 423, "water_hot": 240}
      ]
    }
  }
}

Troubleshooting

Issue	Solution
Wrong numbers	Check meter photo quality, ensure numbers are clear
Can't identify meter type	Name photo file: `electricity.jpg`, `water.jpg`
Vision error	Check MINIMAX_API_KEY is set

AB-Agents 🦀

Requirements

👁️ AB Agents Vision (MiniMax) — Required for image analysis. Install first:

clawhub install AB-Agents-Vision-MiniMax

AB-Agents 🦀

安全使用建议

This skill appears to do what it says (use a vision API to read meter photos and store readings), but there are a few red flags you should address before installing or running it: - It requires a MINIMAX_API_KEY at runtime even though the registry metadata lists no required env vars. Make sure you supply only the expected vision API key and not broader credentials. - The script attempts to source /root/.openclaw/.minimax-env. Inspect that file's contents before running the script (it may contain other tokens or secrets). If you don't control that file or it contains unrelated secrets, do not run the script as root. - The script invokes an external binary 'uvx minimax-coding-plan-mcp' and relies on python3; confirm what 'uvx' is, where it comes from, and that it is the official client for the MiniMax tool you trust. If unsure, run the skill in a sandbox or on a throwaway account. - Because the metadata doesn't declare required env vars or binaries, consider requesting the author to explicitly list MINIMAX_API_KEY and required binaries (uvx, python3) in the skill manifest before use. If you want to proceed safely: inspect meter-reader.sh line-by-line, verify /root/.openclaw/.minimax-env contents (or remove that sourcing line), ensure uvx is trusted, and run the script as an unprivileged user with only the vision API key available in a controlled environment.

功能分析

Type: OpenClaw Skill Name: ab-agents-meter-reader Version: 1.0.2 The skill is a utility for extracting meter readings from images using a Vision LLM via an MCP tool. It processes photos of electricity and water meters, saves the history locally in `~/.meter-readings/`, and generates reports. It uses `uvx` to run the `minimax-coding-plan-mcp` tool and sources environment variables from a specific OpenClaw path (`/root/.openclaw/.minimax-env`), which are standard behaviors within the OpenClaw ecosystem for vision-integrated tasks. No malicious intent or data exfiltration was detected.

能力标签

requires-sensitive-credentials

能力评估

ℹ Purpose & Capability

The name/description (meter reading from photos) matches the code and SKILL.md: the script analyzes images with a MiniMax vision tool, saves history, and formats messages. However, the skill does not declare required runtime artifacts (it uses the 'uvx' tool and python3 and expects a MINIMAX_API_KEY) in the registry metadata, which is inconsistent with its actual needs.

⚠ Instruction Scope

SKILL.md instructs use of MiniMax and installing AB-Agents-Vision-MiniMax, and the script calls a local MCP client ('uvx minimax-coding-plan-mcp') with JSON prompts and images. The script also attempts to source /root/.openclaw/.minimax-env (an absolute path outside the skill's directory) which could cause the agent to read unrelated environment/configuration. The instructions and code rely on MINIMAX_API_KEY but the registry lists no required env vars — this mismatch is significant.

✓ Install Mechanism

No install spec is present (instruction-only + a shell script). That minimizes install-time risk because nothing is automatically downloaded or written during installation. The runtime does call external binaries (uvx, python3) but they are not installed by the skill.

⚠ Credentials

The script requires MINIMAX_API_KEY at runtime (and SKILL.md calls this out), but the skill metadata declares no required env vars. It also sources /root/.openclaw/.minimax-env which may expose other secrets or tokens stored by the platform or other skills. Requesting the vision API key is proportionate to the stated purpose, but silently reading an absolute platform path is not justified by the meter-reading goal.

⚠ Persistence & Privilege

The skill does not request 'always: true' and writes only to a per-user config directory (~/.meter-readings), which is reasonable. The concern is the script's attempt to source a root-level OpenClaw env file (/root/.openclaw/.minimax-env) — that reads platform-level state belonging to the agent environment and could leak unrelated credentials. The skill does not modify other skills but it does access a config path outside its scope.

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 ab-agents-meter-reader
安装完成后，直接呼叫该 Skill 的名称或使用 /ab-agents-meter-reader 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

v1.0.2

- Added a new Requirements section to SKILL.md, specifying AB Agents Vision (MiniMax) as a required dependency for image analysis - Included installation instructions for AB-Agents-Vision-MiniMax using clawhub - No functional or code changes; documentation update only

v1.0.1

- Fixed homepage URL in SKILL.md to point to the correct repository. - No other changes.

v1.0.0

- Initial release of AB-Agents-Meter-Reader. - Reads electricity (single or dual tariff) and water meter readings from photos. - Saves historical readings and tracks multiple apartments. - Generates ready-to-send messages for landlords. - Requires MiniMax Token Plan API key and runs on Linux/macOS.

元数据

Slug ab-agents-meter-reader

版本 1.0.2

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 3

常见问题