← 返回 Skills 市场
ezviz-open

Ezviz Open Safety Production Inspection

作者 EzvizOpenTeam · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ 安全检测通过
181
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install ezviz-open-safety-production-inspection
功能描述
Ezviz safety production inspection skill. Captures device images and sends to Ezviz AI for workplace safety analysis.
使用说明 (SKILL.md)

Ezviz Safety Production Inspection Skill

Captures images from Ezviz cameras and sends to Ezviz AI for workplace safety production inspection.

Template: 安全生产行业通用智能体 (e15f061c13f349b1b2a3)

Detection Items:

  • Safety helmet detection (安全帽佩戴)
  • Fall detection (人员跌倒)
  • Cleanliness analysis (整洁度)
  • Fire/smoke detection (烟火检测)

⚠️ Important: This skill uses the 安全生产行业通用智能体 template.

  • Template ID: e15f061c13f349b1b2a3
  • The skill will prioritize existing agents with '安全生产' in the name
  • If no matching agent exists, a new one will be created from the safety production template
  • Created agents will automatically have '安全生产' in the name

User Confirmation Required

Running this skill means you accept these remote actions:

  1. Query Ezviz agent list (open.ys7.com)
  2. Create agent from template if needed (open.ys7.com)
  3. Capture device images (open.ys7.com)
  4. Send images for AI analysis (aidialoggw.ys7.com)

Data flow: Device -> open.ys7.com -> aidialoggw.ys7.com -> Local output

Privacy:

  • Images stored on Ezviz (2 hours)
  • Token cached in /tmp/ezviz_global_token_cache/ (perms 600)

Security Requirements

  1. Use dedicated Ezviz app credentials (not main account)
  2. Minimal permissions: capture + agent APIs only
  3. Prefer environment variables over config files
  4. Test with non-production devices first
  5. Agent name must contain '安全生产' (compliance requirement)

Quick Start

Environment Variables (Recommended)

export EZVIZ_APP_KEY="your_key"
export EZVIZ_APP_SECRET="your_secret"
export EZVIZ_DEVICE_SERIAL="dev1,dev2"
# Template ID is optional (default: e15f061c13f349b1b2a3)
export EZVIZ_SAFETY_TEMPLATE_ID="e15f061c13f349b1b2a3"
python3 scripts/safety_production_inspection.py

Config File (Alternative)

Add to ~/.openclaw/channels.json:

{
  "channels": {
    "ezviz": {
      "appId": "your_app_id",
      "appSecret": "your_app_secret",
      "devices": ["BF6985110"],
      "safetyTemplateId": "e15f061c13f349b1b2a3"  # 安全生产行业通用智能体模板 ID
    }
  }
}

Then run:

python3 scripts/safety_production_inspection.py

Disable Token Cache (High Security)

export EZVIZ_TOKEN_CACHE=0
python3 scripts/safety_production_inspection.py

Pre-Run Verification

Run these before first use:

# 1. Check for hidden characters
python3 -c "
import sys
for f in ['SKILL.md', 'scripts/safety_production_inspection.py', 'lib/token_manager.py']:
    with open(f, 'rb') as file:
        text = file.read().decode('utf-8')
        for i, c in enumerate(text):
            code = ord(c)
            if code \x3C 32 and code not in [9, 10, 13]:
                print(f'{f}: Control char at {i}')
                sys.exit(1)
print('All files clean')
"

# 2. Verify API domains (should use open.ys7.com only)
grep -r "open.ys7.com" scripts/ lib/  # Should show endpoints

# 3. Verify domain connectivity
curl -I https://open.ys7.com/api/lapp/token/get  # Should return 200 OK

# 4. Check token cache permissions
ls -la /tmp/ezviz_global_token_cache/global_token_cache.json
# Should show: -rw------- (600)

API Endpoints

Domain Purpose
open.ys7.com Token, Capture, Agent Management
aidialoggw.ys7.com AI Analysis

Note: open.ys7.com is the official Ezviz Open API domain (openai = Open API, not AI).


Safety Production Inspection Items

The AI agent will analyze images for:

  1. Personal Protective Equipment (PPE) - Safety helmets, safety vests, goggles
  2. Fire Safety - Fire extinguishers, emergency exits, no smoking signs
  3. Electrical Safety - Proper wiring, no exposed wires, grounded equipment
  4. Workplace Organization - Clear walkways, proper storage, no clutter
  5. Machinery Safety - Guard rails, warning signs, proper operation
  6. Hazardous Materials - Proper storage, labeling, containment

Security Checklist

Before running:

  • Reviewed scripts/safety_production_inspection.py
  • Reviewed lib/token_manager.py
  • Verified API domains (open.ys7.com, aidialoggw.ys7.com)
  • Created dedicated Ezviz app (minimal permissions)
  • Tested with non-production device
  • Configured 安全生产 template ID

Before autonomous use:

  • Accept remote side effects (agent creation, image capture, AI analysis)
  • Understand data flow and privacy implications
  • Configured token cache appropriately for your environment

Update Log

Date Version Changes
2026-03-19 2.0.1 Updated template ID to e15f061c13f349b1b2a3 (安全生产行业通用智能体)
2026-03-19 2.0.0 Renamed from restaurant-inspection to safety-production-inspection
2026-03-19 1.0.9 Agent name must contain '安全生产' (compliance requirement)
2026-03-19 1.0.8 Security hardened: unified domains, user confirmation, clean UTF-8
2026-03-19 1.0.7 Global token cache support
2026-03-19 1.0.6 Config file support

Author: EzvizOpenTeam License: MIT-0

安全使用建议
This skill appears to do what it says: it will authenticate to Ezviz, possibly create or rename an intelligent agent with '安全生产' in the name, capture device images, and send them to Ezviz's AI analysis endpoint (aidialoggw.ys7.com). Before installing or running it: - Use a dedicated Ezviz app key/secret with minimal permissions (capture + agent management) as recommended. - Be aware the script will look for credentials in ~/.openclaw/* if env vars aren't present — remove or audit stored config files if you don't want those credentials used. - The skill caches access tokens under /tmp/ezviz_global_token_cache/global_token_cache.json and sets the file to 0600; review that directory on your host if multiple users share it. Consider disabling caching (EZVIZ_TOKEN_CACHE=0) for higher security. - The skill will send images to aidialoggw.ys7.com and may create agents in your Ezviz account — test with non-production devices and review the two included files (scripts/safety_production_inspection.py and lib/token_manager.py) before execution. If you need higher assurance, ask the author for an explicit code review checklist (particularly for token cache directory permissions and the exact agent creation payload), or run the scripts in an isolated environment (container) and with non-production credentials first.
功能分析
Type: OpenClaw Skill Name: ezviz-open-safety-production-inspection Version: 1.0.0 The skill is a legitimate integration for the Ezviz (萤石) IoT platform, designed to automate workplace safety inspections by capturing camera images and analyzing them via Ezviz AI agents. It communicates exclusively with official Ezviz domains (open.ys7.com and aidialoggw.ys7.com) and implements secure token management using a local cache in /tmp with restricted file permissions (0600). The documentation in SKILL.md is transparent about remote side effects, data flow, and privacy, and the code in scripts/safety_production_inspection.py and lib/token_manager.py contains no evidence of malicious intent or unauthorized data exfiltration.
能力评估
Purpose & Capability
The name/description match the code and runtime instructions: the skill authenticates to Ezviz (open.ys7.com), captures device images, may create/modify intelligent agents, and sends images to aidialoggw.ys7.com for AI analysis. The required environment variables (EZVIZ_APP_KEY, EZVIZ_APP_SECRET, EZVIZ_DEVICE_SERIAL) are appropriate for Ezviz API access.
Instruction Scope
Instructions explicitly describe remote side effects (query agent list, create agent from a template, capture images, send images to aidialoggw.ys7.com). The SKILL.md and scripts also read local OpenClaw config files (~/.openclaw/*) as a credential fallback, and include pre-run verification commands that read the skill files. Those fallbacks are documented in SKILL.md but may be surprising: the skill can proceed using stored credentials from config files even if you did not export env vars.
Install Mechanism
No external install/downloads; this is an instruction-and-script skill that depends on the requests Python package (declared in SKILL.md). There is no arbitrary remote install URL or extracted archive.
Credentials
Requested credentials are proportional to the stated purpose (Ezviz app key/secret and device serial). The skill also supports optional env vars (EZVIZ_TOKEN_CACHE, EZVIZ_SAFETY_TEMPLATE_ID, EZVIZ_CHANNEL_NO) and will fall back to credentials in ~/.openclaw config files. That fallback may cause the skill to use stored credentials without explicit env vars, which users should be aware of.
Persistence & Privilege
The skill does not request always:true and does not modify system-wide agent settings beyond creating/renaming agents in the Ezviz account (an intended remote side effect). It creates a global token cache in /tmp/ezviz_global_token_cache/global_token_cache.json (file permission set to 0600). Token caching is deliberate and documented, but storing tokens in a shared tmp directory is a potential privacy consideration on multi-user hosts.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install ezviz-open-safety-production-inspection
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /ezviz-open-safety-production-inspection 触发
  4. 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
v1.0.0
Ezviz Safety Production Inspection Skill v1.0.0 - Initial release of workplace safety production inspection for Ezviz cameras. - Captures device images and sends them to Ezviz AI for analysis of safety helmet use, fall detection, cleanliness, and fire/smoke hazards. - Supports automatic agent creation based on the "安全生产行业通用智能体" template if needed. - Requires dedicated Ezviz app credentials and device serial(s) via environment variables or config file. - Token securely cached with file permissions set to 600. - Comprehensive user guide and security checklist included.
元数据
Slug ezviz-open-safety-production-inspection
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Ezviz Open Safety Production Inspection 是什么?

Ezviz safety production inspection skill. Captures device images and sends to Ezviz AI for workplace safety analysis. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 181 次。

如何安装 Ezviz Open Safety Production Inspection?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install ezviz-open-safety-production-inspection」即可一键安装,无需额外配置。

Ezviz Open Safety Production Inspection 是免费的吗?

是的,Ezviz Open Safety Production Inspection 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Ezviz Open Safety Production Inspection 支持哪些平台?

Ezviz Open Safety Production Inspection 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Ezviz Open Safety Production Inspection?

由 EzvizOpenTeam(@ezviz-open)开发并维护,当前版本 v1.0.0。

💬 留言讨论