dpia-ai
Skill2 starsChanged 3 months agoApache-2.0View on GitHub
Run or review a GDPR data protection impact assessment for an AI system. Use for Article 35 trigger screening, necessity and proportionality, profiling and automated decisions, bias and explainability risks, mitigations, DPO consultation, Article 36 prior consultation, or DPIA and EU AI Act FRIA coordination.
In this repository5 other files it gives its agents.
- ai-act-classificationSkill
- ai-act-complianceSkill
- ai-vendor-assessmentSkill
- governance-documentationSkill
- risk-managementSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Follow LEGAL-SOURCE-PROTOCOL.md for legal conclusions and source notes.
- Distinctive: line 23
When decisive facts are missing, issue a provisional trigger decision and list the facts needed to finalise it.
- Distinctive: line 27
Document controller, processors, joint controllers, purpose, users, affected persons, data sources and categories, model and system flow, …
- Distinctive: line 31
Test purpose specification, legal basis, data minimisation, accuracy, retention, transparency, rights handling, Article 22 where relevant, …
- Distinctive: line 43
If high residual risk remains in the absence of measures, route to prior consultation under Article 36 before processing.
- Distinctive: line 48
Read AI-ACT-DECISION-MAP.md and validate relevant articles through the Lexbeam EU AI Act MCP.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Run an AI DPIA
Support the controller’s assessment; do not claim approval or legal advice. The controller remains accountable. If a DPO is designated, seek and document the DPO’s advice under GDPR Article 35(2); the GDPR does not make DPO approval a universal legal condition.
Follow LEGAL-SOURCE-PROTOCOL.md for legal conclusions and source notes.
Determine whether a DPIA is required
Apply GDPR Article 35 to the specific processing. Consider:
- systematic and extensive evaluation based on automated processing, including profiling, where decisions have legal or similarly significant effects
- large-scale special-category or criminal-offence data
- systematic monitoring of publicly accessible areas on a large scale
- relevant supervisory-authority lists and EDPB/WP29 guidance
Label the familiar “two or more criteria” approach as guidance, not statutory text. Record the trigger analysis even when the conclusion is that no DPIA is required.
When decisive facts are missing, issue a provisional trigger decision and list the facts needed to finalise it. Do not pause without giving the user the safe analysis available from supplied facts.
Describe the processing
Document controller, processors, joint controllers, purpose, users, affected persons, data sources and categories, model and system flow, recipients, transfers, storage, retention, automated effects, human intervention, and lifecycle changes. Use a data-flow diagram when it clarifies the processing.
Assess necessity and proportionality
Test purpose specification, legal basis, data minimisation, accuracy, retention, transparency, rights handling, Article 22 where relevant, alternatives, human intervention, and processor controls. Do not use an AI Act classification as a GDPR lawful basis.
Assess risks to people
Describe concrete harms, affected groups, likelihood, severity, uncertainty, and controls. Include discrimination, proxy effects, erroneous decisions, opacity, manipulation, security, data leakage, function creep, exclusion, chilling effects, feedback loops, and contestability where relevant.
Distinguish inherent risk, control effectiveness, and residual risk. Link every mitigation to an owner, evidence, due date, and verification method.
Decide and escalate
- Record the DPO advice and the controller’s response where a DPO is designated.
- Consult data subjects or representatives where appropriate under Article 35(9), subject to the stated exceptions.
- If high residual risk remains in the absence of measures, route to prior consultation under Article 36 before processing.
- Set review triggers for purpose, data, model, provider, population, performance, incident, complaint, or legal change.
Coordinate with the EU AI Act
Read AI-ACT-DECISION-MAP.md and validate relevant articles through the Lexbeam EU AI Act MCP. Use provider information under Article 13 for the DPIA where Article 26(9) applies. Assess Article 27 FRIA scope separately. Cross-reference overlapping DPIA sections under Article 27(4); do not say a DPIA automatically replaces a FRIA.
Output
Produce administrative details, trigger decision, system and data-flow description, necessity and proportionality assessment, risk register, measures, residual-risk decision, consultation record, controller decision and governance record, review triggers, evidence list, and source note. Keep DPO advice separate from the controller decision.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

