About this skill

Problem

In private Guanyuan BI instances, the harder issue is often not creating cards, but legacy ETLs with circular references, business rules scattered across calculated columns, metrics inflated by patterns like IS NOT NULL, and engine-level failures in custom charts, v7 page publishing, or SuperApp scaffolding that official DSL cannot reach. This skill targets those practical hard edges with routing rules and failure playbooks.

How it works

It first performs routing: standard querying, card building, ETL, and dataset CRUD go to the official suite such as guancli, guanvis, guanetl, and guands; field-level payloads, engine errors, state-machine bypasses, or reverse engineering then move to the relevant Part. Core capabilities include:
- ETL governance: scan circular dependencies, node complexity, output size, and field references, then produce keep/delete decisions and layering guidance (ODS/DIM/DWD/DWS/APP).
- Full-chain rewrite: convert SmartETL chains into SQL-backed versions, using page-copy acceptance checks, diff localization, and empty-snapshot blocking to reduce false positives.
- Custom charts: troubleshoot payload_json, descriptor, fixed cards/overlays, and existing-page injection issues.
- v7 and SuperApp: cover draft-publish state machines, silent node-level pitfalls, form_xxx table reverse engineering, and LLM response parsing anomalies.

Boundaries

It is a practical enhancement layer over the official tools, not a standalone BI client. Simple read-only queries should route to the official read-only hub; write operations require explicit date ranges, data scopes, and deletion targets. It assumes a local private BI instance, session/profile-based auth, and appropriate write permissions.

Screenshots

Use Cases

Maintain an aging Guanyuan BI setup, scanning ETL cycles, unused nodes, and field references to decide merge or retire tables.
For restaurant BI, encode repeat purchase, new/returning guest, time-slot, and RFM formulas into ETL and dashboards consistently.
Ship multiple HTML app dashboards on v7, bypassing draft state, locating node errors, and validating page diffs.
Debug SuperApp table creation or BI-to-LLM failures by reverse-engineering form APIs and parsing error_message.

Best For

Data engineers owning private Guanyuan BI who need to govern legacy ETLs, split complex nodes, and reduce deletion risk.
Restaurant BI analysts who need to consolidate repeat purchase, guest mix, and RFM definitions into wide tables and dashboards.
App engineers building Guanyuan v7 or SuperApp who must debug page publishing, custom charts, and API anomalies.
Project leads presenting BI rebuild options to leadership who need to decide whether to govern or redesign the base schema.