Praxikon

EU AI Act and GDPR checks

Praxikon checks a concrete use of AI or personal data against the EU AI Act and the GDPR, and shows where every answer comes from. What you can ask:

What you can ask

Is our AI hiring tool high risk under the EU AI Act?
Example path

MCP connector

Server URL: https://www.praxikon.com/api/mcp
Login: None
Protocol: MCP 2025-11-25
Instructions: Read only access to the published Praxikon implementation graph of the EU AI Act, the case check under the GDPR and the AI Act, and the published answers on both. Nothing here is generated: every field comes from a published object or from a deterministic computation over one, and every result carries its sources. To assess one system, call with a coded profile; send a version 2 profile to have the wider set of duties assessed, including Article 26 for a deployer. Use for the legal path behind one obligation and before you quote a statement; both take an identifier from earlier tool output, and without one or is the route. names the supervisor per member state. When the user describes a concrete use of AI or of personal data and asks whether a DPIA, a FRIA or GDPR and AI Act duties apply, call with the coded answers only, never with the case text or personal data; turns the same answers into the DPIA intake as a Word file. For a short general question, call with the question rewritten in general terms, without names or other personal data. If the user asks that Praxikon does not keep their question, pass store_for_improvement false. The prompts dpia-scan, fria-voorbereiding, leverancierscheck-ai and ai-act-classificatie set out fixed workflows. Answers are compact by default; pass detail full only when the user needs the complete objects. Call only when the user asks to talk to a person. This is not legal advice and not a supervisory authority.

Tools

Reads data9

Skills

None in the store listings or on the server

MCP connector

Server URL

https://www.praxikon.com/api/mcp

None

Protocol

MCP 2025-11-25

Instructions

Read only access to the published Praxikon implementation graph of the EU AI Act, the case check under the GDPR and the AI Act, and the published answers on both. Nothing here is generated: every field comes from a published object or from a deterministic computation over one, and every result carries its sources. To assess one system, call with a coded profile; send a version 2 profile to have the wider set of duties assessed, including Article 26 for a deployer. Use for the legal path behind one obligation and before you quote a statement; both take an identifier from earlier tool output, and without one or is the route. names the supervisor per member state. When the user describes a concrete use of AI or of personal data and asks whether a DPIA, a FRIA or GDPR and AI Act duties apply, call with the coded answers only, never with the case text or personal data; turns the same answers into the DPIA intake as a Word file. For a short general question, call with the question rewritten in general terms, without names or other personal data. If the user asks that Praxikon does not keep their question, pass store_for_improvement false. The prompts dpia-scan, fria-voorbereiding, leverancierscheck-ai and ai-act-classificatie set out fixed workflows. Answers are compact by default; pass detail full only when the user needs the complete objects. Call only when the user asks to talk to a person. This is not legal advice and not a supervisory authority.