← 返回 Skills 市场
Performance Monitoring
作者
daxiangnaoyang
· GitHub ↗
· v1.0.0
· MIT-0
85
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install performance-monitoring
功能描述
监控Agent调度、执行及内存性能指标，自动记录日志，分析趋势，识别瓶颈并提供优化建议。
安全使用建议
Before installing, confirm these points with the skill author or maintainer: (1) How will Feishu alerts be authenticated? Demand explicit environment variables (webhook URL or API token) and review them before providing. (2) Exactly which agent logs and 'memories' will be collected, stored, and transmitted? Ask for a data schema, retention policy, and any redaction/filters for sensitive fields. (3) Where are logs stored and who can access them? Verify retention_days and make sure it meets your privacy policy. (4) Because the skill's source and homepage are unknown, prefer running it in a sandboxed or test agent first and require explicit approval before enabling autonomous invocation in production. (5) If you must use it, provide minimal-scoped credentials (dedicated Feishu webhook limited to a test space) and monitor network activity and outputs for unexpected data exfiltration.
功能分析
Type: OpenClaw Skill
Name: performance-monitoring
Version: 1.0.0
The performance-monitoring skill bundle is designed to collect and analyze agent execution metrics such as dispatch success rates, task completion times, and memory efficiency. The provided PowerShell and Python snippets in SKILL.md focus on reading local log files from a specific directory (C:\Users\Administrator\.openclaw\workspace-main\logs) and generating summary reports. No evidence of data exfiltration, unauthorized network access, or malicious intent was found; the code logic is consistent with the stated purpose of system optimization.
能力评估
Purpose & Capability
The name/description (performance monitoring of Agent scheduling, execution, memory) matches the metric calculations and alerting behavior described in SKILL.md and config.json. However, the skill declares no required env vars or credentials while config.json lists alert_channel: "feishu" and SKILL.md mentions recording to 飞书表格 (Feishu). Sending data to Feishu normally requires webhooks/tokens — those are not declared, which is disproportionate to the stated manifest (missing auth requirements).
Instruction Scope
SKILL.md instructs collecting internal agent metrics and 'memories' and recording logs and alerts (including to Feishu). The document implies reading agent-internal data structures (dispatch logs, task logs, memory lists). The instructions (as provided) do not specify what exact data fields will be collected, any filtering/sanitization, or where logs are stored. They also do not explain how Feishu integration is authenticated or whether data is transformed before external transmission. This open-endedness grants the skill broad discretion to read and transmit potentially sensitive agent data.
Install Mechanism
Instruction-only skill with no install spec and no code files — lowest file-based risk. Nothing will be downloaded or written to disk by an installer as part of this package.
Credentials
Requires no environment variables according to the registry, but config and SKILL.md indicate external alerting to Feishu and access to internal memories/logs. External posting to Feishu would normally require a webhook URL or API token (sensitive). The absence of declared credentials or guidance on required secrets is a proportionality mismatch. Also, collecting 'memories' and detailed logs can expose sensitive user data — the skill does not specify scoping or redaction rules.
Persistence & Privilege
always:false (normal) and model invocation not disabled (normal). Autonomous invocation is allowed by default; combined with the above concerns (external posting and broad data access), autonomous runs could exfiltrate data if credentials are provided later by the operator. The skill does not request persistent installation privileges or modify other skills' configs.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install performance-monitoring - 安装完成后，直接呼叫该 Skill 的名称或使用
/performance-monitoring触发 - 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
Performance-monitoring skill v1.0.0
- Initial release with comprehensive agent performance monitoring capabilities.
- Collects, logs, and analyzes key metrics such as scheduling success rate, dispatch time, retry/fallback rates, parallel speedup, task completion, memory count, relevance, retrieval accuracy, and token efficiency.
- Includes alerting on performance anomalies and provides data-driven optimization suggestions.
- Metrics are output in both logs and Feishu tables.
- PowerShell-based implementation with modular metric collection functions.
元数据
常见问题
Performance Monitoring 是什么？
监控Agent调度、执行及内存性能指标，自动记录日志，分析趋势，识别瓶颈并提供优化建议。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 85 次。
如何安装 Performance Monitoring？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install performance-monitoring」即可一键安装，无需额外配置。
Performance Monitoring 是免费的吗？
是的，Performance Monitoring 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Performance Monitoring 支持哪些平台？
Performance Monitoring 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 Performance Monitoring？
由 daxiangnaoyang（@daxiangnaoyang）开发并维护，当前版本 v1.0.0。
推荐 Skills