Loading ToolBench
MCP server for Qlik Sense Enterprise on Windows (client-managed): Repository, Engine/QIX, and Proxy APIs as Model Context Protocol tools for analysts, developers, and admins.
qseow-mcp demonstrates solid naming conventions (verb_noun pattern across all 20 tools) and comprehensive parameter descriptions. However, output schemas are not documented in the source code, and several tools lack sufficient context about return structures. Tool descriptions average ~80 chars, which is below the 194-char baseline for A+ tools. Parameter schemas are present with type definitions, but lack enum constraints for categorical inputs (e.g., visualization types, action types). Error handling is present but generic, no recovery guidance or actionable error messages visible. Security gating for mutations is well-designed (config-driven registration), but no per-tool permission declarations or audit trail patterns evident.
Add a visualization object to a sheet with validation
Create a new application with optional script and reload
Create a hypercube with specified dimensions and measures
Create a new sheet in an application
Cross-check an expression for additive consistency by grouping
Delete an application
List or terminate QPS user sessions
Output schemas not documented. No visible return type definitions for any of the 20 tools. LLMs cannot plan downstream calls or extract required fields without knowing response structure.
Categorical parameters lack enum constraints. 'visualization' in add_sheet_object accepts free-form strings (kpi, barchart, etc.) but no enum defined. 'action' in end_user_sessions (list/terminate) should be an enum. LLMs will hallucinate invalid values.
Tool descriptions are brief (avg ~70 chars vs 194-char baseline). Missing context on WHEN to use each tool and what it returns. E.g., 'Validate a complete sheet for renderability and correctness' does not explain what 'renderability' means or what errors are returned.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 67 | 2026-07-28+ | v2 |
Check Qlik Engine health and resource saturation
Evaluate a Qlik expression and return its value with selection context
Evaluate a master measure by title
Get data lineage for an application (sources and connections)
Get the reload log for a task execution with forensic details
Get the load script for an application
List applications with optional filtering
List reload tasks with optional filtering
Reload an application
Search for expressions, objects, and master items by term
Audit security rules for risky patterns and drift
Set the load script for an application
Validate a complete sheet for renderability and correctness
No error recovery guidance. Error handling is present but generic, no actionable messages like 'Try search_expressions() first' or 'Invalid field: must be one of [Region, Product, Month]'. Agents cannot self-correct.
No per-tool permission declarations or scope hints. Tools like delete_app and set_script are mutation-gated at registration, but no explicit 'requires: write:app' or 'requires: admin' metadata visible. Audit trail patterns absent.