RaptorMCP Knowledge Base
Enterprise Qlik AI Knowledge Base
Your Qlik estate has a working memory.
RaptorMCP keeps the knowledge your Qlik teams rely on in one managed place and makes it available to each user's chosen AI agent, limited to what that user is permitted to see. The agent can then explain, investigate and prepare work using your standards and definitions, not generic guesses.
What the knowledge base holds
The managed knowledge base is organised around how Qlik work is actually done. Content can be linked to Qlik spaces, streams, apps and business areas so the agent finds the right context.
Documents and standards
Development standards, naming conventions, architecture notes, runbooks and handover documents.
Business rules and definitions
Agreed measure definitions, calculation rules, data ownership and the context behind a KPI.
Templates
Load script patterns, app and sheet templates, documentation templates and change checklists.
Extensions and assets
Approved Qlik extensions, themes and binary assets, stored with version history.
Workflow definitions
Approval routes, sign-off steps and escalation rules used by Raptor writeback and workflows.
Reusable agent Skills
Scoped, repeatable methods that guide an AI agent through a Qlik task the way your team does it.
Enterprise, team and user context
Knowledge is organised from enterprise and team standards down to a user's permitted context. Two things decide what is available to a user's agent at any moment:
Enterprise
Organisation-wide standards, definitions and approved Skills that apply to everyone with access.
Team
Team or business-unit folders for domain rules, app families and team-specific methods.
User
A user's own permitted working context, notes and drafts.
- Configured folder permissions set which knowledge folders each user, group or team can read or change.
- Applicable Qlik access limits anything that comes from Qlik itself, such as apps, data and metadata, to what that user may see in Qlik.
Qlik context, Skills and business definitions
When a user asks a question, the agent can combine permitted Qlik context, such as app structure, fields, master items and reload history, with your own business definitions. That is how it can explain not just a number but what the number is meant to measure.
Skills are reusable, scoped methods. An enterprise Skill might describe how your organisation documents an app. A team Skill might describe the checks for a finance reload. Skills are versioned like other knowledge and are only available where they have been enabled.
Versioning, assets and collaboration
- Knowledge is version controlled with Git, so every change has an author, a date and a history you can review or roll back.
- Larger binary assets, such as extensions, images and templates, are stored with Git LFS.
- Several people and agents can work on the same knowledge. Conflict checks stop one change silently overwriting another.
- Captured Qlik scripts are only included where script-read authority has been granted and capture is configured. They are not added to Git by default.
Bringing in existing knowledge
Most teams already have useful knowledge spread across shared drives, wikis and old project folders. Selected material can be imported, but it is planned first:
- Choose what is current and worth keeping, rather than importing everything.
- Decide which enterprise, team or user folder each item belongs in, and who should see it.
- Record provenance: where each item came from and who owns it.
- Review sensitive content before it becomes available to any agent.
Practical examples
Application handover
A developer leaves a project. The knowledge base holds the app's purpose, data sources, model notes and open issues, so the next person's agent can explain the app and its history before anyone changes it.
Measure definition
A business user asks what "net revenue" means in a dashboard. The agent answers from the agreed definition in the knowledge base and points to the expression and app it is used in, within that user's Qlik access.
Repeatable investigation
A reload fails at month end. A team Skill guides the agent through the same checks every time: recent changes, source availability, script errors and data volumes, and records what it found.
Change preparation
Before a script or model change, the agent gathers the relevant standards, dependent apps and previous decisions, then drafts a change note for human review. Making the change is a separate, authorised action.
From knowledge to authorised action
Knowledge informs the agent; it does not grant permission to act. Changing a script, reloading an app, writing back data or starting a workflow are separate actions. Each one needs its own authority in Qlik or in the configured source system, and the Gateway's role and capability settings decide which actions are enabled at all.
Questions and answers
What is an enterprise Qlik AI knowledge base?
It is a managed, central store of the knowledge Qlik teams rely on, such as standards, business definitions, templates, workflow definitions and agent Skills, organised so an AI agent can use it in context. With RaptorMCP it is served to each user's chosen agent through the Gateway.
Who can see what?
Configured folder permissions decide which knowledge each user can reach. Anything that comes from Qlik is further limited by that user's applicable Qlik access. The agent only works with what the user is permitted to see.
Does it include our current Qlik scripts?
Not automatically. Capturing current scripts requires its own script-read authority in Qlik. Where that authority is granted and capture is configured, script snapshots can be versioned alongside other knowledge.
Which AI agent does it work with?
The user's chosen, approved MCP-compatible agent. Data processing for answers depends on the AI model and provider that agent uses and how it is configured. An enterprise-hosted Gateway does not by itself make AI inference local.