grafana-lens
由 awsome-o 提供
DevOps & Cloud简介
Full Grafana access: query, dashboard, alert, trace — 16 agent tools.
描述
该技能的功能
Gives an agent native access to a Grafana instance for querying metrics, logs, and traces, creating and modifying dashboards, managing alert rules, and pushing custom data. Works with any Prometheus, Loki, or Tempo datasource configured in Grafana. Covers security monitoring and distributed trace investigation.
Covers the full observability stack (metrics, logs, and traces) in a single tool chain, so you can go from alert to root cause without leaving the conversation.
何时使用
- Check how much an AI agent spent on LLM tokens today
- Find error logs for a failing API request over the last hour
- Create a dashboard showing LLM latency and cost trends by model
- Alert when daily cloud costs exceed a set threshold
- Debug a slow distributed request by searching traces and correlated logs
示例工作流
User asks: Alert me when my AI agent's daily cost goes over $10
- 1
Calls grafana_explore_datasources to discover the Prometheus datasource UID
- 2
Calls grafana_check_alerts with action 'setup' to create the webhook contact point
- 3
Calls grafana_create_alert with expr 'openclaw_lens_daily_cost_usd', threshold 10, condition 'gt'
- 4
Returns the alert rule URL and confirms metricValidation shows current cost data
Alert rule created. Fires when daily cost exceeds $10; notifications route to the agent via webhook.
前置要求
- Grafana instance URL (grafana.url config)
- Grafana API key with editor or admin permissions (grafana.apiKey config)
安装
clawhub install awsome-o/grafana-lens运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。