Dataworks Smart Monitor
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill’s monitoring purpose is coherent, but it would let a background agent use DataWorks credentials and send potentially sensitive task logs or reports to LLM/Feishu without clear redaction, approval, or recipient controls.
Review carefully before installing. Use only a dedicated least-privilege DataWorks credential, confirm exactly which Feishu channel/webhook will receive reports, require log redaction before LLM or Feishu sharing, and enable cron/sub-agent execution only if you want recurring autonomous monitoring.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Operational logs could contain secrets, customer data, table names, or business details that may be exposed to an LLM provider or Feishu recipients.
The skill says it will collect error logs, send them through LLM analysis, and post reports to Feishu, while acknowledging those logs may contain sensitive information. The artifacts do not specify concrete redaction, approval, recipient, or data-boundary controls.
3. 提取失败任务的错误日志 4. 使用 LLM 分析失败原因 ... 7. 发送报告到飞书 ... 2. **敏感信息** - 错误日志中可能包含敏感信息，注意脱敏
Require explicit redaction and approval before sending logs or reports externally, configure exact Feishu recipients/webhook scope, and document what data is shared.
A broadly scoped DataWorks access key could allow the agent or sub-agent to read more project/job/log data than intended.
DataWorks credentials are expected for this integration, but the registry metadata declares no credentials and the instructions do not define least-privilege permissions or secure storage requirements.
在 `TOOLS.md` 中配置 DataWorks 项目信息： - PROJECT_ID: 33012 - REGION_ID: ap-southeast-1 - ACCESS_KEY_ID: [已配置] - ACCESS_KEY_SECRET: [已配置]
Use a dedicated least-privilege read-only credential, avoid placing long-lived secrets in broadly shared markdown files, and declare the credential requirement clearly.
The monitoring task may continue outside the main conversation or run on a schedule, making API calls and sending alerts without the user watching each step.
The skill intentionally uses a spawned sub-agent and suggests scheduled recurring execution. This is disclosed and aligned with monitoring, but it is autonomous background activity.
sessions_spawn({
agentId: "agent-ge",
task: `分析 DataWorks 任务运行情况（昨天）`,
mode: "run",
streamTo: "parent",
label: "dataworks-smart-monitor"
})
...
建议配置每日上午 9:00 自动执行（使用 cron）Make scheduled execution explicitly opt-in, document how to stop or cancel runs, and limit the sub-agent’s permissions to this monitoring task.
Users may be confused or may obtain/run an unreviewed script from elsewhere to make the README commands work.
The README references Python scripts, but the provided manifest contains only README.md and SKILL.md. Any referenced script behavior is therefore not inspectable in this artifact set.
python3 dataworks_th_smart_monitor.py ... python3 dataworks_th_smart_monitor.py --date 2026-03-01
Package the referenced scripts with the skill or remove the commands; if scripts are external, provide a trusted source and version pinning.