Zworker

控制zworker AI自动化任务应用，支持任务管理、定时计划控制、用户信息同步和消息通知转发。当用户提到“zworker”关键词时使用此技能，例如：同步用户信息、获取任务列表、执行任务、管理定时计划、获取并发送通知等。技能通过HTTP接口与本地运行的zworker应用（localhost:18803）通信。

This skill appears to implement local control of a zworker service and its scripts call only localhost:18803, which is expected. However, the SKILL.md contains a risky, vague instruction to 'obtain userids by any means' and assumes the agent will read OpenClaw's openclaw.json and call platform message tools even though no config paths or permissions are declared. Before installing or enabling this skill: 1) Confirm you trust the local zworker application and that it should receive channel/user mappings. 2) Do not grant the agent broad file-read access — prefer explicitly supplying the openclaw.json path or a concrete users JSON via --file/--users or stdin. 3) If you allow automated runs, restrict the agent's ability to search arbitrary system files. 4) Review the scripts yourself (they are included) and test them in a sandbox: they only make HTTP calls to localhost and do not exfiltrate to remote hosts. 5) Ask the skill author to remove/replace the 'by any means' instruction with a precise, limited method for obtaining userids (e.g., read a configured openclaw.json path) and to declare any required config paths. If you cannot limit file access or get clearer instructions, treat the skill as potentially overbroad and avoid enabling autonomous invocation.

Type: OpenClaw Skill Name: zworker Version: 1.0.3 The 'zworker' skill bundle is a legitimate integration tool designed to interface OpenClaw with a local Electron-based automation application via a localhost HTTP API (http://localhost:18803). The skill facilitates task management, schedule control, and the synchronization of user identifiers (channel and userid) between the two applications. While the instructions in SKILL.md aggressively direct the AI agent to retrieve user IDs from local configuration files, the data is transmitted only to a local endpoint, and the Python scripts (notably scripts/zworker_api.py and scripts/sync_users.py) contain standard, transparent logic without evidence of exfiltration, obfuscation, or malicious intent.