cgi-advantage-reporting

# CGI Advantage Reporting Agent designs CGI Advantage 4, known by the team as Trains 4, reporting flows across native reports, scheduled exports, warehouse extracts, and analytics handoffs. ## When to Use | Condition | Use | |---|---| | Work needs custom financial, operational, or control reports from Trains 4 data | Use this skill | | Work needs scheduling, distribution, or export automation | Use this skill | | Work needs Excel, CSV, PDF, Crystal, Fabric, or Power BI outputs | Use this skill | | Work compares real-time inquiry to staged history | Use this skill | ## When Not to Use | Condition | Route | |---|---| | Work focuses on posting APIs or source-system extraction | Use `cgi-advantage-integration` | | Work focuses on cutover or reconciliation during migration | Use `legacy-trains-migration` | | Work focuses on semantic model design after curated data lands in Fabric | Use `fabric-powerbi-integration` | | Work focuses on generic SQL profiling | Use `data-quality-reporting` | ## Required Inputs | Input | Required | Notes | |---|---|---| | Audience | Yes | Record finance operations, leadership, audit, vendor support, or analytics consumers. | | Grain | Yes | Record transaction, document, vendor, fund, period, or summarized grain. | | Delivery cadence | Yes | Record on-demand, intra-day, nightly, close-cycle, or archival cadence. | | Output format | Yes | Record inquiry, file export, Crystal render, API payload, or semantic dataset. | | Filter set | Yes | Record fiscal period, fund, department, vendor, document, and status filters. | | Source surface | Yes | Record native report, inquiry, report job, staging schema, or warehouse source. | | Volume target | No | Use for row caps, batching, partitioning, and file packaging. | ## Pattern Matrix | Need | Preferred Pattern | Guardrail | Pass Target | |---|---|---|---| | Analyst inquiry or low-latency balance review | Inquiry or API-backed interactive report | Enforce narrow filters, paging, and row caps | Response stays inside the interactive budget | | Large recurring distribution | Scheduled export package | Stamp batch ID, period, row count, and control totals | Export package reconciles to source counts | | Pixel-perfect operational document | Native or Crystal report render | Keep layout logic separate from source-query logic | Rendered output matches approved totals and layout | | Cross-period analytics or dashboard history | Warehouse or lakehouse extract | Land immutable batches with watermark lineage | Load count and watermark match source selection | | Current-period inquiry plus trend history | Hybrid interactive plus batch | Record one authority rule for current values and one for history | Users see no conflicting totals | ## Workflow | Step | Agent Action | Test | Pass | |---|---|---|---| | 1 | Agent maps audience, authority source, cadence, and output contract. | Review the report specification. | Each report has one audience and one authority source. | | 2 | Agent selects interactive, scheduled, rendered, warehouse, or hybrid delivery. | Compare the request to the pattern matrix. | Each report has one primary delivery pattern. | | 3 | Agent defines parameters, joins, export schema, file naming, and control totals. | Review the parameter and output contract. | Filters and reconciliation fields stay explicit. | | 4 | Agent implements query shape, scheduling path, render path, and distribution metadata. | Run a representative render or extract. | Output shape and metadata match the contract. | | 5 | Agent wires downstream Fabric, Power BI, or .NET consumers to the governed report or extract. | Review lineage and refresh hooks. | Each consumer traces to one governed source. | | 6 | Agent validates totals, parameter behavior, large-report performance, and schedule observability. | Execute the verification checklist. | Totals reconcile and runtime stays inside the selected budget. | ## Rules | Topic | Rule | |---|---| | Terminology | Use `Trains 4` in team-facing notes and keep `CGI Advantage 4` in formal references. | | Reuse | Reuse native reports and inquiries before building raw transactional joins. | | Dataset authority | Stabilize the dataset before Crystal or PDF layout work. | | Export packaging | Include period stamp, batch ID, row count, and control totals on every package. | | Analytics lineage | Route Power BI and Fabric loads through governed staged extracts or curated layers. | | Variant consistency | Drive Excel, CSV, PDF, and API variants from one parameter contract. | ## Verification Checklist | Check | Test | Pass | |---|---|---| | Source alignment | Trace one output column set to source fields. | Each output column maps to one documented field or derived rule. | | Parameter behavior | Run narrow and broad filters. | Filters return expected subsets with no ambiguous expansion. | | Export integrity | Generate the required output formats. | Row counts, totals, and file names reconcile across variants. | | Performance | Run one representative high-volume report. | Execution time stays inside the agreed budget. | | Schedule observability | Execute one scheduled run. | Schedule, completion, and retry status remain observable. | | Analytics handoff | Land one extract in staging. | Refresh uses the expected watermark and batch ID. | ## Outputs - Report catalog and audience map - Delivery-pattern decision record - Parameter, filter, and reconciliation contract - Render, extract, and distribution plan - Analytics handoff design - Verification plan for totals, scheduling, and performance ## Reference Files - [Reporting surfaces reference](references/reporting-surfaces.md)