survey-workflow

组织健康度与员工敬业度调研全流程管理 Agent。功能包括：①员工名单批量导入问卷系统；②追加人员（增量导入）；③发送调研通知（支持自定义通知模板）；④追踪填答状态；⑤截止前自动催办；⑥拉取答卷数据（API直连）；⑦计算本批次基准均值；⑧按模板生成部门/集团诊断报告。问卷包含麦肯锡组织健康度37题（10维度）+...

This skill appears to be a legitimate survey workflow integration, but it expects gateway credentials and will process employee identifiers and submission data. Before installing: 1) Confirm the API endpoints belong to your organization (production domains shown are org-specific). 2) Do not provide a high-privilege org-wide appKey or long-lived access-token; instead create a scoped service account / token that only permits the needed operations (import targets, sendNotify, list submissions, pressure) and can be revoked. 3) Update the skill manifest to declare required env vars (e.g., APP_KEY, ACCESS_TOKEN) so the platform can surface and control secrets. 4) Test against the provided test environment before hitting production. 5) Decide and document where local archives (input/notification_records.json) are stored and who can read them; avoid storing tokens in plaintext. 6) Audit usage: monitor notification and pressure endpoints carefully (pressureNotify is a GET that triggers side effects) to avoid accidental spam or unauthorized notifications. If you cannot supply minimal, scoped credentials or cannot verify the service domains, treat this skill as not ready for deployment.

Type: OpenClaw Skill Name: survey-workflow Version: 1.0.0 The survey-workflow skill bundle manages employee engagement surveys by executing local Python scripts and interacting with external APIs (mediportal.com.cn and xgjktech.com.cn). While its behavior is aligned with the stated purpose, it possesses high-risk capabilities including the handling of sensitive employee PII (IDs and names), network access for data exfiltration to external endpoints, and the execution of shell commands for report generation. Under the provided criteria, these risky capabilities are considered suspicious even if plausibly needed for the workflow, as they represent a significant attack surface without further sandboxing of the scripts (e.g., generate_dept_reports_v4.py) and API interactions.