About this skill

Problem

Frontend performance issues often appear across pages, APIs, static resources, and exception logs. A dashboard may show that LCP is slow, error rates are rising, or API latency is increasing, but it does not always explain whether the cause is page loading, resource requests, JavaScript errors, business retcode anomalies, or regional network differences. This skill focuses on Tencent Cloud RUM and turns metric queries, log retrieval, and result synthesis into a reviewable analysis with evidence and actionable optimization suggestions.

How it works

It first uses QueryRumWebProjects to confirm the target application, then selects the right tool for the request. Exception analysis uses the exception metric: JavaScript errors can be narrowed with level=4, while JavaScript and Promise errors can use level in ('4','8'). API latency, HTTP status codes, and retcode behavior are analyzed through network. Page performance defaults to LCP; pv, uv, and resource cover traffic and static-resource loading. Queries follow RUM parameter constraints: GroupBy must be an array, Filters must be a JSON object, and multi-dimensional analysis should be split into separate queries to avoid Cartesian-product row explosions. Metric queries default to Limit 100, while log queries default to Limit 10. URL details in logs are usually embedded in the msg field and should be filtered with like. When logs contain trace data, the skill can also connect to APM for deeper link analysis.

Boundaries

It fits Web frontend performance, API stability, static resources, and exception logging in Tencent Cloud RUM. It is not intended for pure backend service performance, native mobile app performance, non-Tencent Cloud RUM platforms, or general coding tasks. A valid RUM credential is required. If the ProjectId or application name is uncertain, the correct application must be selected first. Empty results may require widening the time range, and is_err filters only retcode errors rather than HTTP status-code errors.

Screenshots

Use Cases

After a release, page LCP regresses and the team needs to trace slow frontend loading by page URL and region.
After many JavaScript or Promise errors appear in production, query RUM exception logs and summarize frequent root causes.
When API latency rises or retcode anomaly rates increase, analyze top slow and error endpoints by interface URL.
When users report slow asset loading, review static-resource metrics under a page to identify high-latency resources.

Best For

Frontend monitoring engineers who need to trace LCP, error rates, and API latency to specific pages, endpoints, or regions.
Engineering leads conducting release reviews who need RUM metrics and logs to evidence before-and-after performance changes.
Frontend engineers debugging online JavaScript/Promise errors who need frequent error context from exception logs.
Product engineers focused on web experience who need PV/UV and resource-loading metrics to identify bottlenecks.