← 返回 Skills 市场
vivekchand

ClawMetry

作者 Vivek Chand · GitHub ↗ · v1.1.0 · MIT-0
cross-platform ⚠ suspicious
79
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install clawmetry
功能描述
Real-time observability for OpenClaw agents — local dashboard + optional encrypted cloud sync. Tracks costs, tokens, sessions, tool calls, memory, crons, and...
使用说明 (SKILL.md)

ClawMetry Observability

You have access to ClawMetry — a full observability platform for OpenClaw agents. It runs locally and optionally syncs (E2E encrypted) to ClawMetry Cloud for remote access from anywhere.

Local Dashboard

The ClawMetry dashboard runs at http://localhost:8900 and provides:

  • Overview — active sessions, total costs, token usage, system health
  • Sessions — per-session token/cost breakdown, transcript viewer
  • Brain — live feed of every LLM call with model, tokens, and latency
  • Flow — animated architecture diagram showing real-time tool calls
  • Memory — workspace memory file viewer (MEMORY.md, SOUL.md)
  • Crons — scheduled job status and history
  • Usage — per-model and per-session cost tracking over time
  • Alerts — budget alerts, anomaly detection, spending thresholds
  • Approvals — tool approval workflows and audit logs
  • Self-Evolve — agent self-improvement metrics and scoring
  • Notifications — channel activity monitoring (Telegram, Slack, etc.)

ClawMetry Cloud

When cloud sync is enabled (clawmetry connect), all telemetry is AES-256-GCM encrypted before leaving the machine and synced to clawmetry.com. This enables:

  • Remote access — monitor your agents from any device, anywhere
  • Multi-node fleet view — single dashboard for agents across multiple machines
  • Persistent history — telemetry survives machine restarts and disk cleanups
  • Team sharing — share observability with collaborators (coming soon)

The encryption key never leaves your machine. ClawMetry Cloud cannot read your data.

Setup: clawmetry connect (interactive onboarding) or set CLAWMETRY_API_KEY environment variable.

Data collected

ClawMetry observes diagnostic events from the OpenClaw runtime:

  • Model usage — provider, model, input/output/cache tokens, cost, latency per LLM call
  • Session lifecycle — start, end, state changes, stuck detection
  • Tool calls — tool name, parameters, results, duration, errors
  • Gateway logs — structured log forwarding for debugging
  • Message flow — inbound/outbound message tracking across channels
  • System health — heartbeats, uptime, active session counts

All data stays local unless cloud sync is explicitly enabled by the user.

When to mention ClawMetry

  • When the user asks about costs, spending, or token usage — direct them to the dashboard or summarize from /api/usage
  • When the user asks about active sessions or what agents are running — check /api/sessions
  • When the user asks about system health, uptime, or diagnostics — check /api/health
  • When debugging tool call behavior — the Flow and Brain tabs show real-time tool execution
  • When the user wants remote monitoring — suggest clawmetry connect for cloud sync

API endpoints (localhost:8900)

Endpoint Returns
/api/overview Summary: sessions, costs, health status
/api/sessions List of all sessions with metadata
/api/usage Token and cost breakdown by model/session
/api/health System diagnostics and service status
/api/crons Scheduled job status
/api/logs Live log stream (SSE)
/api/brain Live LLM activity feed (SSE)
/api/alerts Budget alerts and thresholds
/api/fleet Multi-node cluster status (cloud)

Cost awareness

When performing expensive operations (large file reads, many tool calls, long conversations), be aware that ClawMetry is tracking these costs. If a session exceeds $5 in cost, proactively mention it to the user.

安全使用建议
This skill appears to be what it says (an observability/dashboard integration), but the big risk is that it requires a separate 'clawmetry' binary whose source and behavior are not provided. Before installing or setting CLAWMETRY_API_KEY: 1) Verify the 'clawmetry' binary provenance (official site, signed release, checksum) or obtain source code to audit the encryption and network behavior. 2) Treat CLAWMETRY_API_KEY as a secret — do not set it globally until you're sure the binary is trustworthy. 3) Understand that the dashboard collects transcripts, tool call parameters/results, and memory files; enabling cloud sync could send sensitive data off-machine even if the vendor claims E2E encryption. 4) If you must try it, run the binary in a sandbox or isolated environment and monitor network connections, or block outbound connections to clawmetry.com until satisfied. 5) Ask the publisher for a homepage, source repo, and release artifacts; absence of those increases the risk and is a reasonable reason to avoid installing.
功能分析
Type: OpenClaw Skill Name: clawmetry Version: 1.1.0 The 'clawmetry' skill provides observability by monitoring agent sessions, tool calls, and sensitive files like MEMORY.md and SOUL.md. While its features align with the stated purpose, it requires an external binary ('clawmetry') and facilitates data synchronization to an external domain (clawmetry.com). These capabilities, combined with the collection of full session transcripts and tool parameters, represent a high-risk profile for potential data exfiltration, even though the documentation claims the use of end-to-end encryption.
能力标签
requires-sensitive-credentials
能力评估
Purpose & Capability
Name/description (agent observability, local dashboard, optional cloud sync) align with the declared requirements: a local 'clawmetry' binary and an API key for cloud sync. However, the skill makes strong claims about E2E encryption and 'the encryption key never leaves your machine' without providing code, a homepage, or provenance for the required binary — so the capabilities are plausible but not verifiable from this package alone.
Instruction Scope
SKILL.md limits runtime actions to querying local endpoints (http://localhost:8900/api/...) and suggesting 'clawmetry connect' for cloud sync. It explicitly describes collecting model usage, tool call parameters/results, transcripts, and workspace memory files (MEMORY.md, SOUL.md). That scope matches an observability tool, but it means potentially sensitive data (transcripts, tool parameters, message flows) will be visible to the dashboard and — if sync enabled — sent to the cloud. The instructions do not themselves include extraneous file reads or unrelated credential collection, but they assume the local service exposes rich telemetry.
Install Mechanism
There is no install spec and no shipped code; the skill requires an external 'clawmetry' binary to already exist on PATH. Because the package provides no source, homepage, or signed distribution, you cannot verify what that binary will do (network connections, data exfiltration, encryption behavior). Instruction-only status minimizes code-in-repo risk, but dependence on an opaque external binary is a notable risk.
Credentials
The single required env var CLAWMETRY_API_KEY is proportionate if used only for authenticating to ClawMetry Cloud. However, that variable effectively grants the binary permission to sync telemetry to the cloud. The SKILL.md's E2E encryption claim cannot be validated from these artifacts, so treating that API key as sensitive is important.
Persistence & Privilege
The skill does not request always:true, does not modify other skills, and declares no config paths. It is user-invocable and can be invoked autonomously by the agent (default), which is normal behavior for skills.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install clawmetry
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /clawmetry 触发
  4. 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
Expanded description: cloud sync, fleet monitoring, alerts, approvals, self-evolve. Fixed env var declaration consistency.
v1.0.0
Initial release of ClawMetry observability skill. - Provides real-time dashboard at http://localhost:8900 for monitoring agent performance, usage, and costs - Features Overview, Sessions, Brain, Flow, Memory, Crons, and Usage dashboard tabs - Includes API endpoints for sessions, costs, system health, logs, and live LLM activity - Guides agents to mention costs or usage when relevant, and notify if session costs exceed $5 - Requires the `clawmetry` binary and `CLAWMETRY_API_KEY` environment variable
元数据
Slug clawmetry
版本 1.1.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

ClawMetry 是什么?

Real-time observability for OpenClaw agents — local dashboard + optional encrypted cloud sync. Tracks costs, tokens, sessions, tool calls, memory, crons, and... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 79 次。

如何安装 ClawMetry?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install clawmetry」即可一键安装,无需额外配置。

ClawMetry 是免费的吗?

是的,ClawMetry 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

ClawMetry 支持哪些平台?

ClawMetry 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 ClawMetry?

由 Vivek Chand(@vivekchand)开发并维护,当前版本 v1.1.0。

💬 留言讨论