qlik-toolkit

Designs Qlik Sense data architecture: app architecture (single app, generator/consumer, four-layer split, binary load), star schema, ETL layer boundaries, QVD layer design, cross-layer field mapping, key resolution, source-architecture consumption (warehouse, OLTP, Data Vault 2.0, flat files), incremental load strategy, and master calendar requirements. Use this agent for from-scratch data model design, choosing app architecture for volume/refresh/team constraints, diagnosing synthetic keys or grain mismatches, or adapting a source architecture to Qlik consumption. See "When to invoke" in the agent body for triggers.

Generates project documentation from existing Qlik artifacts. Produces any of nine documents (README, data dictionary, technical specification, expression catalog, visualization guide, deployment runbook, user guide, change log, dependency tracker). Audience-calibrates: technical content for developers, plain language for business users. Use this agent for end-of-project release documentation, refreshing a single doc after a change, producing a stakeholder-specific document standalone, or capturing blocked dependencies for handover. See "When to invoke" in the agent body for triggers.

Authors Qlik Sense expressions: master measures, master dimensions, calculated dimensions, set analysis expressions, variable expressions, and complex aggregations. Produces an expression catalog and a runnable expression-variables.qvs file when scope warrants. Use when writing or reviewing Qlik expressions, whether one-off or as a full catalog. Iterative by design — comfortable filling gaps or fixing issues as they emerge. See "When to invoke" in the agent body for triggers.

Reviews Qlik Sense development artifacts (data models, load scripts, expressions, viz specs, full apps) against best practices, naming conventions, expression correctness, performance patterns, and cross-artifact consistency. Performs post-load data quality validation against a live tenant when MCP access is available. Read-only by design — produces severity-rated findings with remediation guidance, doesn't fix issues. Use this agent for a focused review of a single artifact, a comprehensive cross-artifact QA pass, verifying fixes from a prior review, or running post-load data quality checks. See "When to invoke" in the agent body for triggers.

Conducts structured discovery for Qlik Sense projects. Two capabilities, usable independently or together: platform discovery for brownfield work (cataloging existing apps, subroutine libraries, naming conventions, QVD layouts, and platform constraints) and business requirements gathering (user personas, source systems, business rules with grain, ETL preferences, refresh, and security needs). Use this agent at project start, when inheriting an existing Qlik platform, or whenever a structured discovery pass is needed before downstream design begins. See "When to invoke" in the agent body for triggers.

Post-load data quality validation patterns for Qlik Sense. Provides query templates for null-rate analysis, referential integrity checks, value distribution analysis, row-count validation, orphaned-record detection, sparse-field identification, and duplicate-key detection — usable against a live tenant via MCP or any post-load data interface. Also provides patterns for embedding validation checks directly into load scripts so failures surface during reload rather than in production. Load when running data quality checks against a loaded app, writing diagnostic or validation scripts, or designing fail-fast assertions inside a load script.

Capability registry for the Qlik Cloud MCP server: maps MCP tools to development tasks, provides MCP detection patterns, documents behavioral gotchas not covered by tool definitions, and defines multi-step workflows for expression validation, reference app analysis, visualization scaffolding, and data quality checks against a live tenant. The MCP tool definitions themselves document parameters and response structure; this skill covers framework integration, sequencing, and pitfalls discovered through live testing. Load when interacting with a live Qlik Cloud tenant via MCP, validating an expression against real data, analyzing a reference app, or running post-load data quality checks.

Star schema design, key resolution (natural/composite/hash/AutoNumber), synthetic key prevention, circular reference resolution, QVD layer architecture, multi-app patterns (single app, generator/consumer, extract-transform-model-UI, binary load), source architecture consumption (dimensional warehouse, OLTP, Data Vault 2.0, pre-joined views, flat files), grain alignment across multiple facts, bridge tables, link tables, ApplyMap vs dimension, HidePrefix/HideSuffix. Load when designing or reviewing a Qlik Sense data model, deciding between single-app and multi-app, choosing a key strategy, or diagnosing synthetic keys and circular references.

Set analysis syntax and patterns ({1}, {$}, element-set arithmetic, P()/E(), $(=...) modifiers), aggregation functions, TOTAL qualifier usage, Aggr() patterns, conditional expressions, null handling in expressions, dollar-sign expansion timing, expression performance optimization, calculation conditions, and common Qlik Sense expression anti-patterns. Load when writing, reviewing, or debugging Qlik expressions — master measures, master dimensions, calculated dimensions, KPI formulas, set-analysis snippets, year-over-year and rolling-window calculations, or any expression that misbehaves under selection.

Script syntax reference, QVD optimization, incremental load patterns (insert-only, insert/update, insert/update/delete, dual-timestamp for SCD2), JOIN/KEEP prefixes, ApplyMap patterns, CROSSTABLE, master calendar generation, variable definitions, error handling, logging patterns, null handling patterns, diagnostic and validation patterns, subroutine integration, and platform gotchas (SET vs LET, dollar-sign expansion timing, SET variable comma limitation). Load when writing, reviewing, or debugging Qlik load scripts, QVD operations, STORE/LOAD syntax, preceding LOAD, NullAsValue, script organization, JOIN, KEEP, ApplyMap, CROSSTABLE, AutoNumber, composite keys, or data quality defensive coding.

1 hook across 1 event