Assess obligations under Regulation (EU) 2023/2854 for connected products, related services, statutory B2B data sharing, unfair data-contract terms, exceptional-need B2G requests, data-processing-service switching, international governmental access, interoperability, smart contracts, enforcement, and non-EU representatives. Use for Data Act scope analysis, gap assessments, product and service design, request handling, contract review, cloud exit planning, or cross-regulation mapping.
Perform a source-grounded assessment under Regulation (EU) 2023/2854. Separate enacted legal requirements from guidance, national law, contractual choices, and implementation recommendations.
This skill is not legal advice. Escalate unresolved interpretation, national enforcement, trade-secret refusal, personal-data processing, and authority-response decisions to qualified counsel.
Source and Evidence Protocol
Start with the current consolidated text on EUR-Lex. Record the ELI/CELEX identifier, consolidation date, language, and access date.
Cite the controlling article and paragraph for each legal conclusion. Read the full provision before asserting that a duty, exception, right, or remedy does not exist.
Label every proposition as one of:
Law: binding text in Regulation (EU) 2023/2854 or another identified legal act.
Official guidance: Commission or competent-authority material, not the Regulation itself.
National overlay: applicable national legislation, authority designation, procedure, or penalty.
Implementation recommendation: operational advice, not a statutory requirement.
Open question: missing fact or unresolved interpretation.
Never infer a legal rule from a heading, recital, search miss, summary page, draft bill, or template alone.
Quote sparingly. Prefer an accurate paraphrase tied to the exact provision.
Commission Data Act page: https://digital-strategy.ec.europa.eu/en/policies/data-act
Commission register of competent authorities and national penalty measures, where available
Controlling Dates
Event
Date
Authority
Entry into force
11 January 2024
Article 50
General application
12 September 2025
Article 50
Article 3(1) design obligation for connected products and related services placed on the market after this date
12 September 2026
Article 50
Chapter IV applies to qualifying legacy contracts
12 September 2027
Article 50
Switching charges prohibited
12 January 2027
Article 29(1)
Do not rewrite the 12 September 2026 rule as a product-design date. It is tied to when the connected product or related service is placed on the market.
Assessment Workflow
Follow the steps in order. Stop and identify missing evidence instead of filling factual gaps with assumptions.
1. Fix the unit of analysis
Identify:
legal entity and establishment;
product, related service, data-processing service, contract, or request being assessed;
relevant market and user location;
assessment date and legal-text version;
data flows, contractual chain, technical control, and actual data availability.
Do not assess an enterprise in the abstract when different products or services create different roles.
2. Assign roles using Article 2 and Article 1(3)
Assess each role independently:
Role
Anchor
Core test
User
Article 2(12)
Owns the connected product, has temporary contractual use rights, or receives related services
Data holder
Article 2(13)
Has a legal right or obligation to use and make data available
Data recipient
Article 2(14)
Receives data in a professional capacity and is not the user
Provider of a related service
Articles 2(6), 3(3)
Provides qualifying software or digital service linked to product functions
Provider of a data processing service
Article 2(8)
Provides qualifying on-demand network access to scalable and elastic computing resources
Public sector body
Article 2(28)
Qualifying Member State authority, public-law body, or association
Smart-contract vendor or deployer for others
Articles 1(3)(g), 36
Supplies or professionally deploys a smart contract used to execute a data-sharing agreement
The Regulation does not define “manufacturer” in Article 2. For product-facing duties, identify who designs or manufactures the connected product and who is the seller, rentor, lessor, or related-service provider under Article 3.
Apply territorial scope from Article 1(3). Non-EU establishment does not by itself remove an entity from scope. Check the legal-representative duty in Article 37(11) to (13).
Read references/scope-assessment.md.
3. Classify the data and applicable chapter
Record:
personal, non-personal, or mixed data;
product data under Article 2(15), related service data under Article 2(16), and relevant metadata;
readily available data under Article 2(17);
content versus data concerning performance, use, or environment;
raw or pre-processed data versus inferred or derived information;
trade-secret status and identity of the trade-secret holder;
security, intellectual-property, confidentiality, and sector restrictions.
Chapter II covers product and related-service data within Article 1(2)(a), not every record associated with a product. Do not assume all analytics or inferred information are readily available data.
For personal data, the GDPR and ePrivacy rules continue to apply and prevail in a conflict under Article 1(5). A Data Act access right is not itself a universal GDPR lawful basis.
4. Test the Chapter II product and service duties
Access by design and pre-contract information
Article 3(1): design and provide connected products and related services so relevant data and metadata are, by default, easily, securely, and free of charge available in the required format and, where relevant and technically feasible, directly accessible.
Article 3(2): seller, rentor, or lessor gives the listed information before a purchase, rent, or lease contract.
Article 3(3): related-service provider gives the listed information before the related-service contract.
User access
If direct access is unavailable, Article 4(1) requires the data holder, on a simple request where technically feasible, to make readily available data and relevant metadata accessible without undue delay, at the same quality available to the holder, easily, securely, free of charge, and in a comprehensive, structured, commonly used, machine-readable format. Continuous and real-time access applies only where relevant and technically feasible.
User-directed third-party sharing
Article 5(1) requires comparable access for an eligible third party requested by the user, subject to Articles 8 and 9. A DMA gatekeeper is not an eligible third party under Article 5(3). Article 6 governs how the recipient may use and further disclose the data.
Exceptions and safeguards
Security restrictions: Article 4(2), with the required competent-authority notification.
User trade secrets: Article 4(6) to (9).
Third-party trade secrets: Article 5(9) to (12).
Personal data: Articles 4(12) and 5(7).
Chapter II enterprise exception: Article 7(1). This is tied to data generated by products manufactured or designed, or related services provided, by qualifying micro or small enterprises, subject to linked-enterprise and subcontracting conditions. It is not a general exemption for every small data holder.
Read references/data-access-rights.md.
5. Test B2B data-sharing conditions
Where a data holder must make data available to a business recipient under Article 5 or other qualifying law:
use fair, reasonable, non-discriminatory, and transparent terms under Article 8;
do not impose contractual terms that Article 8(2), Article 12(2), or Article 13 makes non-binding;
assess compensation under Article 9, not under a fabricated frequent-or-complex-request rule;
for an SME or qualifying not-for-profit research recipient without non-SME linked or partner enterprises, compensation cannot exceed the Article 9(2)(a) making-available costs;
provide the calculation basis in sufficient detail under Article 9(7).
Article 9 governs B2B compensation between data holder and data recipient. It does not permit charging the user for Article 4 or Article 5 access.
6. Review unilaterally imposed B2B data terms
Article 13 applies to specified data-access, data-use, liability, and remedy terms unilaterally imposed by one enterprise on another.
Use the correct structure:
Article 13(1): an unfair, unilaterally imposed term is not binding.
Article 13(2): terms reflecting mandatory or default Union law are not unfair.
Article 13(3): general unfairness standard.
Article 13(4): terms that are unfair.
Article 13(5): terms presumed unfair.
Article 13(6): unilateral-imposition test and burden of proof.
Article 13(7) to (9): severability, exclusions, and non-derogation.
There is no Article 13 safe harbor. Commission model terms under Article 41 are non-binding.
Read references/unfair-terms-catalogue.md.
7. Validate exceptional-need B2G requests
Use the sequence in Articles 14 to 22:
Article 14: confirm the requester and data holder are covered.
Article 15: identify the exceptional need.
Article 17: validate the request’s content, specificity, proportionality, legal-task basis, deadline, and publication/notification steps.
Article 18: calculate the response or challenge deadline and test only the listed grounds to decline or seek modification.
Articles 19 and 21: control use, security, erasure, trade secrets, and permitted onward sharing.
Article 20: determine compensation.
For a public emergency, qualifying non-micro and non-small data holders provide data free of charge under Article 20(1). Micro and small enterprises can claim compensation through Article 20(3). For Article 15(1)(b) requests, Article 20(2) generally allows technical and organisational costs plus a reasonable margin, subject to the official-statistics exception in Article 20(4).
Read references/b2g-data-sharing.md.
8. Assess data-processing-service switching
For services within Article 2(8) and Article 1(3)(f):
Articles 23 and 24: identify covered services and obstacles;
Article 25: check written contract terms, maximum two-month notice, maximum 30-calendar-day transition, exceptions, minimum 30-calendar-day retrieval period, erasure, exportable-data categories, and assistance;
Article 26: check switching information and online register;
Article 27: check good-faith cooperation;
Article 28: check website disclosure on international access and transfer;
Article 29: apply the switching-charge timeline;
Article 30: assess technical duties by service type;
Article 31: test custom-built and non-production exceptions;
Article 32: assess third-country governmental access to non-personal data held in the Union.
Article 24 does not set a notice period. The maximum two-month notice and the transition rules are in Article 25.
Read references/cloud-switching.md.
9. Assess smart contracts
Article 36 applies to smart contracts used to execute data-sharing agreements. Check:
robustness and access control;
safe termination and interruption;
data archiving and continuity;
governance-layer and smart-contract-layer access control;
consistency with the data-sharing agreement;
conformity assessment and EU declaration of conformity.
Do not cite Articles 30 or 31 for these duties.
10. Map enforcement and remedies
Article 10: certified dispute settlement for specified disputes. Decisions bind only with prior explicit consent.
Article 37: competent authorities, data coordinator, jurisdiction, legal representative, and authority powers.
Article 38: complaint.
Article 39: effective judicial remedy.
Article 40: Member State penalties and specified GDPR-linked fine powers.
Article 41: non-binding Commission model terms and cloud clauses.
Verify current authority designations and national penalties from official sources at the assessment date. Do not state a draft bill as enacted law.
11. Integrate adjacent regimes
Map, do not merge:
GDPR and ePrivacy for personal data and terminal-equipment access;
sector data-access law and Article 43 priority rules;
Trade Secrets Directive and Data Act-specific safeguards;
Digital Markets Act gatekeeper status;
Cyber Resilience Act and sector security law;
AI Act where data feeds an AI system;
consumer, competition, intellectual-property, employment, and secrecy law;
Data Governance Act where an intermediary or data-altruism structure is used.
Check current consolidated texts and application dates for every adjacent regime. Read references/cross-regulation-mapping.md and references/dach-specific.md.
Mandatory Output
Deliver one assessment with these sections:
Executive conclusion: in scope, out of scope, or unresolved, with the three most material risks.
Facts and assumptions: entity, product/service, data, geography, contracts, and missing facts.
Role and chapter matrix: each role, factual basis, article, confidence, and evidence.
Requirement matrix: provision, binding rule, present state, evidence, gap, owner, due date, and confidence.
Request or contract findings: exact clause/request, legal test, result, and remediation.
Cross-regulation conflicts: GDPR and other overlays, with separate legal bases.
Action plan: prioritized remediation, validation step, and decision owner.
Source register: official URL, instrument/version, provision, access date, and proposition supported.
Use confidence labels:
High: controlling text and material facts are clear.
Medium: law is clear but a material fact or national overlay is incomplete.
Low: interpretation, scope, or evidence is genuinely unresolved.
Use references/templates.md for reusable matrices and response structures.
Quality Gate
Before finalizing, confirm:
Every role uses the correct Article 2 definition or a clearly described functional duty.
Article 3(1), (2), and (3) are not reversed.
Article 7 is not stated as a general small-data-holder exemption.
Article 9 compensation is not based on request frequency or complexity.
Article 13 uses paragraphs 3, 4, and 5 correctly and no safe harbor is claimed.
B2G grounds, response deadlines, and compensation match Articles 15, 18, and 20.
Cloud notice, transition, retrieval, and charge dates match Articles 25 and 29.
Smart-contract duties cite Article 36.
Authorities, complaints, remedies, and representatives cite Articles 37, 38, and 39 correctly.
Personal-data processing has a separately identified GDPR basis.
Negative legal claims are based on the full provision, not a failed search.
National-law and guidance statements have current official sources and dates.
Articles 3 to 12, access, sharing, safeguards, compensation
references/unfair-terms-catalogue.md
Article 13 contract review
references/b2g-data-sharing.md
Articles 14 to 22 request review
references/cloud-switching.md
Articles 23 to 32 cloud switching and international access
references/cross-regulation-mapping.md
GDPR and adjacent EU-law mapping
references/dach-specific.md
Germany, Austria, and Switzerland verification framework
references/templates.md
Assessment, request, contract, and source-register templates
Disclaimer
This skill supports structured compliance analysis. It does not determine disputed facts, replace qualified legal advice, or guarantee an authority or court outcome. Obtain legal review before refusing access, disclosing trade secrets or personal data, responding to an authority, launching a regulated product, or relying on a national-law position.
Creates an Agent Authority Charter for enterprise or regulated AI agents before deployment. Use this Skill when a user needs to define what an AI agent is allowed to do, who delegated authority to it, what actions are permitted or prohibited, when human approval is required, what evidence must be preserved, and how the agent can be suspended, revoked, or escalated.
EU AI Act Knowledge Engine — authoritative regulatory Q&A grounded in 70 official EU source documents (including the 2026 Commission draft guidelines on Art. 6 high-risk classification — general principles, Annex I, Annex III). Answers any question about the AI Act with article-level citations from the full regulation text, Commission guidelines, EDPB/EDPS opinions, codes of practice, harmonised standards, FRIA guides, incident templates, and sector-specific guidance. This skill should be used when the user asks about the "EU AI Act", "AI Act regulation", "Regulation 2024/1689", "KI-Verordnung", asks to "explain an AI Act article", "what does Article X say", "AI Act requirements", "AI Act penalties", "AI Act timeline", "GPAI obligations", "high-risk AI", "prohibited AI practices", "AI Act and GDPR", "fundamental rights impact assessment", "AI Act standards", "AI Act national implementation", "AI literacy", "serious incident reporting", "Digital Omnibus", "AI regulatory sandbox", or any question about EU artificial intelligence regulation. Also triggers on German terms: "KI-Gesetz", "KI-Verordnung", "Hochrisiko-KI", "KI-Kompetenz", "Grundrechte-Folgenabschaetzung", "GPAI-Verhaltenskodex". For assessment workflows that produce a classification decision rather than a knowledge answer, offer to run a structured risk-tier classification in-session.
This skill builds a structured audit trail of an AI-assisted task: what the tool was asked to do, what materials it was given, what it produced, how the output was verified, and what was ultimately relied upon. It documents the workflow for supervision and later review without ruling on privilege or disclosure, so the record stays accurate rather than self-serving.
Use this skill when the user wants an AI governance, legal-risk, privacy, compliance, procurement, or vendor-risk review of an internal AI use case, an AI product feature, an LLM workflow, or a third-party AI vendor. The skill asks intake and clarifying questions first when facts or evidence are missing, identifies required documentation and missing evidence, maps the use case to AI governance frameworks and applicable legal domains, and produces a preliminary or final governance review with scorecards, findings, owners, remediation actions, and follow-up questions.
Turn an interpretive-ambiguity audit of a legal text — contract, statute, regulation, or judicial opinion — into a polished deliverable. Produces a multi-page website (default), a single-page interactive site, a Microsoft Word document, a PowerPoint deck, or a LaTeX/PDF. Use whenever a user has a stress-test result, ambiguity audit, or "where will this be litigated" scenarios and wants to publish, present, or share them. Triggers on "publish the audit," "make a report from this," "turn this into a website / deck / brief / Word doc," "make a Netlify site," "build slides of these ambiguities," "render the scenarios," "package this stress test," "produce a LaTeX version," and similar. Pairs with the ambiguity-stress-test skill (the underlying analysis) but works on any structured or semi-structured ambiguity analysis. Visual design is opinionated but the user can override colors, fonts, and brand label. Most formats need Python and a writable filesystem; single-file HTML works where the host has neither.
Adversarially stress-tests a legal text — a contract, statute, regulation, or judicial opinion — for interpretive ambiguity: it finds the seams where the people governed by it will later disagree about what it means and turns each into a concrete dispute scenario with both sides' arguments, the likely outcome, and a fix. Use it whenever someone wants to pressure-test, red-team, audit, or find weak spots, loopholes, gaps, ambiguities, or drafting problems in a legal document; whenever a drafter wants to tighten a contract, statute, regulation, or opinion before it issues; whenever a litigator wants to mine an opinion or contract for arguments; or whenever someone hands over a legal text and asks where it will be fought over or for issue-spotting. Trigger it for contract review, statutory-ambiguity analysis, judicial-opinion scope analysis, and drafting QA — even if the user never says "stress-test" or "ambiguity."