Suitable for agent tasks, automation jobs, model services, and support queues. Without runtime logs, do not claim measured uptime, latency, or real cost.
Execution Highlights
Light limestone canvas, white content areas, indigo-purple selected states, with a small amount of green marking confirmed health. Prioritize the task list, status filters, and detail area, then supplement with throughput, duration, and failure trends. All statuses come with text; retries are not counted again as successful tasks, and the definition of whether the success rate denominator includes success/failure/cancellation should be explicit. Latency percentiles are based on individual request records; do not average multiple p95s. Cost descriptions specify currency, pricing date, and usage basis; if price or token records are missing, hide cost estimates. Error aggregation retains the most recent occurrence time, count, and affected tasks, and task results can be expanded for verification. The native dashboard provides a read-only monitoring view and does not draw retry, pause, or delete buttons that do not exist. Primary color #6058DE, background #F1F3F5, body text #293046.
Conversation Example
When asked "see what's wrong," first check the error rate, backlog, duration, and recent status changes, and sort by actual severity of anomalies; when there is no evidence of an anomaly, state the observation scope and do not fabricate failures.
How DB-GPT Works
- Communicate in the user's language. First infer intent from the selected data, prior context, dashboard, and annotations; when the expression is vague but actionable, state one reasonable assumption and continue. Only ask one specific question when a gap would change the result, and do not repeat generic clarification phrases.
- Selected databases are used for read-only discovery only; confirm fields, granularity, units, date ranges, null values, and unique keys. When there is no data, layout can still be discussed; before generating actual metrics, ask the user to select a data source. Illustrative values must be clearly labeled and must never be disguised as query results.
- When creating an editable dashboard, call the existing dashboard tool and load the dashboard-builder workflow: first plan_dashboard, and create according to the existing confirmation process; when modifying an existing dashboard, first load_dashboard_draft, then propose changes matching the version. Visual specifications supplement the business plan and must not skip SQL validation, save, or publish rules.
- Only use the report generation tool when the user explicitly requests an HTML report; the native dashboard must not be secretly replaced with an image or standalone HTML. The native schema uses only the fields, chart types, and theme attributes allowed by the tool, and does not pass arbitrary CSS or nonexistent layout enums. Decorations that cannot be fully expressed should be reduced to usable color schemes/layouts, with the differences explained.
- Before editing, preserve the user's data, metric definitions, and unmodified components. Metric changes must synchronize titles, units, and calculation definitions; for ratio aggregation, sum numerators and denominators first. Do not fabricate targets, year-over-year comparisons, or forecasts when there is no target data.
- Before delivery, check empty results, long labels, narrow screens, light and dark themes, legends, zero values, and missing values; animations should respond to the user's reduced-motion preference. Report only execution results confirmed by the tool.
Resources and Boundaries
This directory is a DB-GPT product skill; it does not modify the host's global configuration, install external scripts, or send data to reference websites. It uses fonts already provided by the system and project, and falls back to Chinese fonts when external fonts are missing. The interface shows a style demonstration; business values come from the selected data during actual use.
References
Reference Sive's related skills scenario and visual direction, with rules rewritten by DB-GPT Dashboard Lab; not an official Sive release package.