rpe-grafana

What This Skill Does

The rpe-grafana skill enables the OpenClaw AI agent to interface directly with your Grafana instances to retrieve real-time data visualizations without requiring knowledge of the underlying database architecture. By abstracting away complex languages like PromQL or SQL, the skill allows you to query metrics, sensor readings, and system statistics simply by referencing dashboard and panel names. It functions by programmatically fetching the panel configuration from Grafana and executing the existing data retrieval logic, effectively 'viewing' the dashboard as a human would, but extracting the raw values for machine-readable use.

Installation

To integrate this skill, ensure your OpenClaw environment is active and install the package using the command line: clawhub install openclaw/skills/skills/darkstards9/rpe-grafana. Once installed, you must configure authentication. You can either update your openclaw.json configuration file under the plugins.entries.rpe-grafana object with your Grafana URL, username, and password, or set these as environment variables (GRAFANA_URL, GRAFANA_USER, GRAFANA_PASSWORD). A user with 'Viewer' level access is sufficient, adhering to the principle of least privilege.

Use Cases

This skill is ideal for teams relying on Grafana for operational visibility. It excels in automated reporting, where you might ask the agent to summarize system health across multiple dashboards during a morning brief. It is also highly effective for incident response; instead of manually navigating through web dashboards during an outage, an engineer can query the agent to provide the current CPU, memory, or error rate values from specific panels instantly. It acts as a bridge between your monitoring stack and your natural language interface.

Example Prompts

1. "Check the 'Production Kubernetes' dashboard and tell me the current CPU usage percentage from the 'Cluster Load' panel."
2. "What are the latest sensor readings for the manufacturing unit dashboard?"
3. "List all available dashboards, then get the value of the 'Total Requests' panel from the API Performance dashboard."

Tips & Limitations

For optimal performance, ensure your dashboard panels are labeled clearly, as the agent relies on these names to discover metrics. Note that this skill is strictly for read-only operations. It cannot modify dashboards, adjust alert rules, or perform administrative tasks. Furthermore, it is not a general-purpose query engine; it requires a pre-configured panel to function. If you need to perform raw database queries not captured by a visual panel, you should use native database tools instead.