← 返回 Skills 市场
smyx-sunjinhui

Safe Driving Behavior Analyzer | 安全驾驶行为分析工具

作者 smyx-sunjinhui · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
69
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install smyx-drive-analysis
功能描述
Analyzes videos of vehicle drivers to identify unsafe driving behaviors. It generates professional analysis reports to help enhance road safety awareness. |...
安全使用建议
Before installing or running this skill: - Treat it as a networked analysis tool: uploaded videos and metadata are sent to remote API endpoints (ApiEnum.BASE_URL_* configured in YAML/config). Do not upload private/sensitive videos unless you trust the destination and have audited the endpoints. - The package reads/writes local config files and uses a local SQLite DB under the workspace 'data' directory—expect persistent storage. If you need no local persistence, do not run it as-is. - The skill bundle uses environment variables (OPENCLAW_WORKSPACE, OPENCLAW_SENDER_OPEN_ID, FEISHU_OPEN_ID) and reads config.yaml files for open-id / api-key, but the registry shows no declared env requirements. Audit or sandbox the runtime environment and avoid providing credentials you care about. - There is no install spec despite a long requirements.txt; you'll likely need to manually install Python dependencies before running. Review requirements and install inside an isolated virtualenv/container. - The repository contains an unrelated face-analysis sub-skill and large common utilities. Ask the author for a minimal build that contains only the drive_analysis pieces or request a clear description of where analysis data is sent (full API hostnames). - If you decide to proceed, run it in an isolated environment (container or VM), inspect util.RequestUtil/http functions to confirm the exact remote endpoints, and avoid supplying real personal identifiers or API keys until you verify the service operator and privacy policy.
功能分析
Type: OpenClaw Skill Name: smyx-drive-analysis Version: 1.0.0 The skill bundle exhibits high-risk behaviors, most notably in `smyx_common/scripts/skill.py`, which uses `subprocess` to recursively call the `openclaw` agent CLI, potentially allowing for uncontrolled agent actions. The `RequestUtil` class in `smyx_common/scripts/util.py` automatically attempts to register or log in the user by sending their `open-id` or `username` to remote endpoints (e.g., `lifeemergence.com`) and stores resulting session tokens in a local SQLite database (`smyx-common-claw.db`). Furthermore, `SKILL.md` contains 'Forced Memory Rules' that instruct the AI to ignore its standard local and long-term memory (LanceDB) in favor of the author's API, which is a significant behavioral override. The bundle also contains a complete nested secondary skill (`face_analysis`), which is unusual and complicates the security audit.
能力标签
requires-sensitive-credentials
能力评估
Purpose & Capability
The code implements video upload, remote API calls, and formatting of analysis reports which align with the described driving analysis purpose. However the package also bundles an unrelated 'face_analysis' sub-skill (中医面诊) and a large shared 'smyx_common' utility set. Many common utilities and a local SQLite DAO exist that go beyond a minimal 'drive-analysis' tool; this is plausible (shared libraries) but increases the surface area and should be expected by the operator.
Instruction Scope
SKILL.md explicitly forbids reading local 'memory' files and LanceDB for historical reports and mandates that history come from cloud APIs. Despite that, the codebase reads/writes YAML config files (skills/smyx_common/scripts/config.yaml and per-skill config.yaml), uses a local SQLite DB (dao.py writing under a workspace 'data' directory), and BaseEnum.YamlUtil will create config files if missing. The skill also instructs the agent to save uploaded attachments to an attachments directory. Those local file I/O behaviors partially contradict the 'do not read local memory' rule and expand persistence beyond what the SKILL.md emphasizes.
Install Mechanism
The skill bundle includes many Python source files and large requirements lists (skills/smyx_common/requirements.txt, skills/face_analysis/requirements.txt) but provides no install spec. That mismatch means the code expects dependencies yet there is no declared install mechanism—this is an incoherence (either dependencies won't be installed or manual install is required). The requirements list is large and includes networking and security-sensitive packages (requests, httpx, openai, cryptography, etc.).
Credentials
Registry metadata lists no required environment variables, but the code reads several environment variables and configuration locations: OPENCLAW_WORKSPACE, OPENCLAW_SENDER_OPEN_ID / OPENCLAW_SENDER_USERNAME, FEISHU_OPEN_ID, plus it loads skill-local config.yaml files to obtain api-key/open-id and ApiEnum base URLs. The SKILL.md mandates a strict open-id retrieval flow that reads local config files; requiring open-id makes sense, but the skill also will consult environment variables for defaults without declaring them. This mismatch (undeclared env usage + optional API keys) is notable and potentially privacy-sensitive because videos and metadata are uploaded to remote API endpoints.
Persistence & Privilege
The code creates/uses local files: it may create missing config.yaml files (YamlUtil.load writes defaults), writes attachments to a skill attachments directory, and uses a local SQLite DB under a workspace 'data' directory. The skill does not set always:true, nor does it modify other skills' configurations, but it does persist data to disk and maintain a local DB—so it gains persistent local presence and storage beyond a pure in-memory analysis script.
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 smyx-drive-analysis
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /smyx-drive-analysis 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
Initial release of "drive-analysis" skill. - Provides professional video-based analysis of vehicle driver behaviors to identify unsafe driving, including distraction, fatigue, and improper seatbelt use. - Strict rules to always retrieve report history from the cloud API, never from local or long-term memory. - Requires a valid open-id (from config or user input) before running any analysis or queries. - Automatically triggers analysis on video/file upload and report listing on specific keywords. - Generates structured analysis reports with safety recommendations; outputs history as a Markdown table with direct report links.
元数据
Slug smyx-drive-analysis
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

Safe Driving Behavior Analyzer | 安全驾驶行为分析工具 是什么?

Analyzes videos of vehicle drivers to identify unsafe driving behaviors. It generates professional analysis reports to help enhance road safety awareness. |... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 69 次。

如何安装 Safe Driving Behavior Analyzer | 安全驾驶行为分析工具?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install smyx-drive-analysis」即可一键安装,无需额外配置。

Safe Driving Behavior Analyzer | 安全驾驶行为分析工具 是免费的吗?

是的,Safe Driving Behavior Analyzer | 安全驾驶行为分析工具 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Safe Driving Behavior Analyzer | 安全驾驶行为分析工具 支持哪些平台?

Safe Driving Behavior Analyzer | 安全驾驶行为分析工具 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 Safe Driving Behavior Analyzer | 安全驾驶行为分析工具?

由 smyx-sunjinhui(@smyx-sunjinhui)开发并维护,当前版本 v1.0.0。

💬 留言讨论