qlik-toolkit

May 30, 2026 · View on GitHub

Make Claude Code fluent in Qlik Sense. Install once and Claude becomes a Qlik developer you can delegate to. It brings deep platform knowledge, specialist agents for different tasks, and an automatic syntax check on every script.

What This Gives You

Claude actually understands Qlik. Without this plugin, Claude drifts into SQL syntax when writing Qlik load scripts, mishandles set analysis, and confuses chart-context behavior with script-context behavior. With it, Claude stops making those mistakes.
Specialist agents on demand. Different jobs need different focus: data architecture, scripting, expressions, visualization, QA review. The right agent shows up automatically based on what you describe. No forced pipeline, no slash commands to memorize.
A syntax safety net. When Claude writes or edits a .qvs file, common mistakes are flagged immediately so you catch them before reload, not after.

Quick Start

Install from the Claude Code desktop UI

In the left sidebar under Personal plugins, click the + button.
Choose Create plugin → Add marketplace.
In the URL field, enter Pupfish-LLC/claude-plugins, then click Sync.
Once the marketplace is synced, click the + button again and choose Browse plugins.
Find Qlik toolkit and install it. The toggle next to the plugin name enables it.

Install from the command line

Add the Pupfish plugin marketplace (if you haven't already):
```
/plugin marketplace add Pupfish-LLC/claude-plugins
```
Install the plugin:
```
/plugin install qlik-toolkit@pupfish
```

Then use it

Just describe what you're doing in Qlik. Examples:

"Help me design a data model for this sales dataset" → loads qlik-data-modeling and offers to invoke the data-architect agent.
"Review my load script for issues" → loads qlik-review-checklist and invokes qa-reviewer.
"What's the difference between SET and LET?" → loads qlik-load-script, answers from skill content.
"Write a set analysis expression for YoY sales growth" → loads qlik-expressions, returns an expression with correct alternate-state and selection-scope handling.

There are no slash commands to memorize. The skills auto-load when their descriptions match your prompt.

What's Inside

Skills (12)

Skill	What it covers
qlik-load-script	Script syntax reference, QVD optimization, incremental load patterns (insert-only, insert/update, insert/update/delete, SCD2 dual-timestamp), `JOIN`/`KEEP`, `ApplyMap`, `CROSSTABLE`, master calendar, error handling, null handling, diagnostic patterns.
qlik-data-modeling	Star schema, key resolution (natural / composite / hash / AutoNumber), synthetic-key vs circular-reference distinction, QVD layer architecture, multi-app patterns (single, generator/consumer, four-layer, binary load), source-architecture consumption (dimensional warehouse, OLTP, Data Vault 2.0, pre-joined views, flat files), grain alignment across multiple facts.
qlik-expressions	Set analysis syntax and modifiers, `TOTAL` qualifier, `Aggr()` patterns, conditional expressions, null handling in expressions, dollar-sign expansion, calculation conditions, common anti-patterns.
qlik-performance	Memory optimization (field types, dual values, symbolic keys), script load optimization (optimized QVD load rules, redundant disk reads), expression calculation optimization, calculation conditions, data reduction techniques, profiling and diagnostic approaches.
qlik-visualization	Chart type selection guide, layout patterns, color and formatting, filter design, responsive design, accessibility, reference app reverse-engineering, Dashboard Bundle controls (Variable Input pipe pattern).
qlik-naming-conventions	Field, variable, table, expression, and file naming standards. Entity-prefix dot notation. Cross-layer field mapping from source through ETL layers to UI display. Reserved words and character restrictions.
qlik-cloud-mcp	Capability registry for the Qlik Cloud MCP server. Tool-to-task mapping, MCP detection patterns, behavioral gotchas not covered by the tool definitions, multi-step workflows (expression validation, reference app analysis, viz scaffolding, data quality checks).
qlik-review-checklist	Complete QA checklist used by the `qa-reviewer` agent: data model integrity, naming compliance, script quality, expression correctness, cross-artifact consistency, blocked dependency audit, data quality validation.
data-quality-validator	Post-load data quality validation query templates: null rate analysis, referential integrity, value distribution, row count validation, orphaned record detection, sparse field identification, duplicate detection, and patterns for embedding the same checks directly into load scripts.
source-profiler	Source data profiling: query templates for source schemas, field types, cardinality, null rates, sample values, data quality indicators. Includes source architecture classification (Dimensional Warehouse / OLTP / Data Vault 2.0 / Pre-Joined Views / Flat Files / API) with consumption implications per type.
qlik-platform-discovery	Brownfield Qlik platform patterns: subroutine identification and limitation patterns from existing `.qvs` scripts, naming-convention variations, connection conventions, QVD storage conventions, architecture pattern detection, convention-conflict resolution when platform conventions differ from framework defaults. Bundles a Platform Context Document template.
qlik-project-scaffold	Cross-platform Qlik project directory scaffolder. Creates a minimal, unopinionated baseline structure (data-sources, scripts, qvds, documentation, tests) with starter READMEs. Idempotent (safe to re-run).

Agents (7)

Think of these as a specialist team you can call on at any point:

data-architect: Designs Qlik data models. Star schema, key strategy, QVD layer architecture, ETL boundaries, source-architecture consumption patterns. Works from a project description, source schema, ERD, existing app, or a conversational ask.
script-developer: Writes production-grade Qlik load scripts (.qvs). Extraction, transformation, QVD generation, incremental loads, master calendar, error handling, diagnostics. Comfortable with reload-feedback fix cycles.
expression-developer: Authors Qlik expressions — set analysis, master measures and dimensions, calculation conditions, variable definitions. Iterative: ranges from one-off snippet to full catalog.
viz-architect: Designs sheet layouts, chart selections, filter panes, and navigation flow. Surfaces expression gaps when visualizations need measures that don't yet exist.
qa-reviewer: Reviews any Qlik artifact or combination (data model, scripts, expressions, full app) against quality standards. Produces structured findings with severity and remediation guidance. Read-only by design.
requirements-analyst: Conducts structured discovery — business requirements, platform context for brownfield projects, user personas, business rules with grain.
doc-writer: Generates project documentation from whatever Qlik artifacts exist. Audience-calibrated (technical for developers, plain language for business users).

Hook

PostToolUse on Write/Edit: Runs validate-qvs-syntax.sh on any .qvs file written or edited. Scans for SQL constructs in LOAD context (HAVING, Count(*), IS NULL, BETWEEN, IN, CASE WHEN, LIMIT, SELECT DISTINCT), TRACE statements with embedded semicolons (which terminate the statement early), unbalanced control blocks (IF/END IF, SUB/END SUB, FOR/NEXT), and malformed PurgeChar() calls (missing second argument). Findings appear before you reload.

1. Ad-hoc syntax help. You: "What's the difference between SET and LET in Qlik?" Claude (with qlik-load-script auto-loaded): Explains that SET preserves the right side as literal text (template), LET evaluates immediately. Notes the gotcha that SET HidePrefix=Chr(37); assigns the literal string, not %.

2. Data model design. You: "I have a sales fact with customer, product, and store dimensions. Help me model it." Claude (with qlik-data-modeling auto-loaded): Walks through key resolution strategy, suggests entity-prefix naming, explains synthetic-key prevention. If the design gets complex, Claude may offer to invoke the data-architect agent for a structured design pass.

3. QA review pass. You: "Review this load script for issues." (pastes script) Claude (with qlik-review-checklist auto-loaded): Invokes the qa-reviewer agent. Returns findings classified as Critical / Warning / Suggestion with file locations, what's wrong, impact, and recommended fixes.

4. Set analysis help. You: "Write a set analysis expression for YoY sales growth." Claude (with qlik-expressions auto-loaded): Returns Sum({<[Cal.FiscalYear]={$(=Max([Cal.FiscalYear]))}>} [Order.Amount]) - Sum({<[Cal.FiscalYear]={$(=Max([Cal.FiscalYear])-1)}>} [Order.Amount]) and explains the alternate-state handling.

What's NOT In This Version

Two planned skills are still in development and will land in a future release:

qlik-security: Section Access patterns for row-level and column-level data security. Until it ships, refer to the Qlik Cloud Section Access docs.
qlik-deploy: App deployment patterns: data connections, reload tasks, space management, and environment promotion. Until it ships, refer to Managing apps in Qlik Cloud.

Why No Top-Level Coordinator?

The plugin intentionally ships without a top-level agent that would run the other agents through a fixed sequence.

Reason: Claude already routes naturally based on what you describe. Each skill's description is the trigger. Each agent declares its inputs and capability. A coordinator layer would sit between you and Claude's routing, and the cost is flexibility. The intended use cases for this plugin are broad: a quick syntax check, a one-off review, a full project build, a brownfield code audit, a single expression tweak. A fixed sequence doesn't fit all of those; a fluid prompt-driven flow does.

If a use case emerges where a coordinator would clearly help, a slash command can be added in a later release. For now, agents and skills, invoked by intent.

Roadmap

Ship qlik-security (Section Access patterns).
Ship qlik-deploy (app deployment patterns).
Optional slash commands for the highest-traffic workflows (driven by user feedback).

Known Out-of-Scope Areas

Topics this plugin intentionally does not cover in v1.0. Each is referenced elsewhere in the plugin (in skill bodies or agent guidance) and pointed back to this list.

Section Access (row-level / column-level security). Deferred to v2.0 pending a full rewrite against current Qlik Cloud documentation. The original draft contained foundational inaccuracies. Until then, refer to the Qlik Cloud Section Access docs. Referenced from the "What's NOT In This Version" section above and from the qa-reviewer agent's security-gap notes.
MCP connections to non-Qlik databases. The qlik-cloud-mcp skill covers the Qlik Cloud MCP server only. Profiling and querying external sources (Redshift, Snowflake, SQL Server, Postgres, etc.) via their own MCP connections is delegated to source-profiler Path B (manual profile completion when a live MCP connection isn't available). Referenced from qlik-cloud-mcp/SKILL.md and source-profiler/SKILL.md.

Feedback

Issues, suggestions, and contributions: please open an issue at github.com/Pupfish-LLC/qlik-toolkit/issues.

License

See LICENSE.

About

Built by Pupfish Analytics. Part of the Pupfish Claude Marketplace. The patterns in this plugin are drawn from years of Qlik consulting engagements, distilled into reusable Claude Code skills that any Qlik developer can install and benefit from. All examples in this release use a generic sales / retail domain.