← 返回 Skills 市场
856
总下载
0
收藏
2
当前安装
3
版本数
在 OpenClaw 中安装
/install arc-skill-health-monitor
功能描述
Monitor deployed skills for performance drift, errors, and unexpected behavior changes. Continuous post-deployment health checks with alerting and trend trac...
使用说明 (SKILL.md)
Skill Health Monitor
Catch skill degradation before it becomes a crisis. Monitors response times, error rates, output drift, and resource usage for deployed skills.
Why This Exists
Skills work fine during testing, then silently degrade in production. Free models change behavior, APIs add latency, memory leaks accumulate. By the time you notice, your agent has been running on broken skills for hours.
Commands
Monitor a skill execution
python3 {baseDir}/scripts/health_monitor.py check --skill \x3Cname> --cmd "python3 path/to/script.py"
View health dashboard
python3 {baseDir}/scripts/health_monitor.py dashboard
Set alert thresholds
python3 {baseDir}/scripts/health_monitor.py threshold --skill \x3Cname> --max-latency 5000 --max-errors 3
Export health report
python3 {baseDir}/scripts/health_monitor.py report --json
View trends for a skill
python3 {baseDir}/scripts/health_monitor.py trend --skill \x3Cname> --period 24h
What It Tracks
- Latency: Execution time per invocation, p50/p95/p99 percentiles
- Error rate: Failed executions, error types, frequency
- Output drift: Detects when output format or content changes unexpectedly
- Resource usage: Memory and CPU at execution time
- Uptime: Availability over time windows (1h, 24h, 7d)
Alerting
- Console alerts when thresholds are exceeded
- JSON webhook support for external integrations
- Configurable per-skill thresholds
Data Storage
Health data is stored in ~/.openclaw/health/ as JSON files. One file per skill, rotated daily.
安全使用建议
This skill is largely coherent and local-only, but review these points before installing:
- Feature mismatch: SKILL.md mentions webhook alerting and daily rotation; the shipped script does not implement webhooks nor daily file rotation — alerts are printed to the console and health is stored in a single per-skill JSON file. If you need webhooks or rotation, do not assume they exist.
- Command execution: The tool runs commands you pass it (intended to run skill entrypoints). The script blocks many shell metacharacters and uses subprocess.run(shell=False), which reduces shell-injection risk, but you should only monitor commands you trust. Avoid wrapping untrusted strings in the --cmd argument.
- Local storage: Health data is stored under ~/.openclaw/health as JSON. Data written there includes truncated error messages and an output 'hash' (Python's built-in hash is not stable across interpreter runs), so output-drift detection may be flaky across restarts.
- If you require webhook notifications, finer retention policies, or stronger drift-detection, inspect and modify the script yourself or request an updated version implementing those features.
Overall: safe to run for trusted uses, but treat it as a local utility and verify the missing features and command usage semantics before relying on it in production.
功能分析
Type: OpenClaw Skill
Name: arc-skill-health-monitor
Version: 1.1.0
The OpenClaw skill bundle 'arc-skill-health-monitor' is classified as benign. The `SKILL.md` provides clear, non-malicious instructions. The core script `scripts/health_monitor.py` handles user-provided commands (`--cmd`) with robust security measures, including input validation (`_sanitize_skill_name`, `_validate_cmd`) and, most importantly, executing external commands via `subprocess.run` with `shell=False`. This critical safeguard prevents shell injection vulnerabilities. Data storage is confined to a standard local directory (`~/.openclaw/health/`), and there is no evidence of data exfiltration, persistence mechanisms, or prompt injection attempts against the AI agent.
能力评估
Purpose & Capability
Name/description align with the shipped Python script: the tool runs a skill command, records latency/error info, stores JSON under ~/.openclaw/health, and provides dashboard/report/trend functionality. Required binary (python3) is proportional.
Instruction Scope
SKILL.md advertises JSON webhook alerting and daily rotation of files, but the script only prints console alerts and writes a single per-skill JSON file (truncating to last 1000 checks). The script does execute arbitrary commands provided by the user (intended for monitoring), and it applies a character-level check to reject shell metacharacters and uses subprocess.run(shell=False). That is reasonable, but the documentation overpromises features that are not present in code.
Install Mechanism
No install spec (instruction-only) and the only required binary is python3. A lightweight single Python script is included; nothing is downloaded or executed from remote URLs.
Credentials
The skill requests no environment variables or external credentials. It only reads/writes files under the user's home (~/.openclaw/health). That is proportional to its stated purpose.
Persistence & Privilege
Does not request persistent platform privileges (always:false). It writes only to ~/.openclaw/health and does not modify other skills or global agent settings.
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 arc-skill-health-monitor - 安装完成后，直接呼叫该 Skill 的名称或使用
/arc-skill-health-monitor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
v1.1.0
- Initial release of version 1.1.0.
- Updated scripts/health_monitor.py.
- No user-facing changes to documentation.
v1.0.1
- Minor changes and improvements to scripts/health_monitor.py.
- No updates to documentation or user-facing commands.
- No change to features or behavior described in SKILL.md.
v1.0.0
Initial release of Skill Health Monitor.
- Continuously monitors deployed skills for latency, error rates, output drift, resource usage, and uptime.
- Provides command-line tools for health checks, dashboard viewing, alert threshold configuration, exporting reports, and viewing trends.
- Alerts via console and JSON webhooks when thresholds are exceeded.
- Stores health data per skill as daily JSON files for trend tracking and reporting.
元数据
常见问题
Skill Health Monitor 是什么？
Monitor deployed skills for performance drift, errors, and unexpected behavior changes. Continuous post-deployment health checks with alerting and trend trac... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 856 次。
如何安装 Skill Health Monitor？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install arc-skill-health-monitor」即可一键安装，无需额外配置。
Skill Health Monitor 是免费的吗？
是的，Skill Health Monitor 完全免费（开源免费），可自由下载、安装和使用。
Skill Health Monitor 支持哪些平台？
Skill Health Monitor 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（darwin, linux）。
谁开发了 Skill Health Monitor？
由 ArcSelf（@trypto1019）开发并维护，当前版本 v1.1.0。
推荐 Skills