00 INDEX

Skill ABQ-Institute/euai_act-compliance-skills/EUAIA/00_INDEX

EU AI Act compliance skills for AI agents — role-aware, risk-tiered, LLM-agnostic. CC BY 4.0.From the repository description

Install

npx -y skills add ABQ-Institute/euai_act-compliance-skills --skill 00_INDEX

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.9 KB, ~1.5k tokens by cl100k_base, as published. Nobody here has run it

EUAIA SKILLS FOLDER — MASTER ROUTER

What this folder is

This folder contains compliance skills derived from Regulation (EU) 2024/1689 (EU Artificial Intelligence Act). These skills are LLM-agnostic and orchestrator-agnostic, designed to be injected into any AI agent operating within or on behalf of an organisation subject to the EUAIA.

Full applicability: 2 August 2026. Prohibited practices: live from 2 February 2025. GPAI obligations: live from 2 August 2025.

TIER A — ALWAYS INJECT AT SESSION START (every agent, every session)

LOAD ALWAYS: 00_INDEX/role_classifier.md 00_INDEX/risk_tier_classifier.md 01_CORE/core_1_prohibited_practices.md 01_CORE/core_2_transparency_baseline.md 01_CORE/core_3_human_oversight.md 01_CORE/core_4_fundamental_rights.md

Target total Tier A token budget: <= 2,500 tokens. If token budget is critically constrained: core_1 is last to be cut. Never cut core_1.

TIER B — LOAD ON DEMAND (RAG or orchestrator-triggered)

IF role_declaration.provider == true LOAD: 02_ROLE_SKILLS/provider/ (all files)

IF role_declaration.deployer == true LOAD: 02_ROLE_SKILLS/deployer/ (all files)

IF role_declaration.importer == true LOAD: 02_ROLE_SKILLS/importer_distributor/imp_1_importer_duties.md 02_ROLE_SKILLS/importer_distributor/imp_3_third_country_providers.md

IF role_declaration.distributor == true LOAD: 02_ROLE_SKILLS/importer_distributor/imp_2_distributor_duties.md

IF risk_tier == HIGH_RISK LOAD: 03_HIGH_RISK/ (all files)

IF risk_tier == PROHIBITED LOAD: core_1 (already Tier A) — HARD STOP — do not proceed

IF system is or uses a GPAI model LOAD: 04_GPAI/ (all files)

IF agent is user-facing LOAD: 05_TRANSPARENCY/ (all files)

IF agent has operational/production scope LOAD: 06_INCIDENTS_AND_MONITORING/ (all files)

IF context contains sector keywords (see map below) LOAD: matching skill from 07_SECTOR_SKILLS/

IF agent scope includes compliance reporting or sanctions LOAD: 08_SANCTIONS/ (all files)

IF context is R&D, testing, sandbox, or pre-market LOAD: 09_INNOVATION/ (all files)

IF a HUMAN is in the conversation and needs to understand their role obligations LOAD: 11_ROLE_GUIDANCE/00_ROLE_GUIDANCE_ROUTER.md then the relevant role guidance file as directed by the router

SECTOR KEYWORD TRIGGER MAP

"patient","clinical","diagnosis","medical","hospital","pharmaceutical","triage" -> sector_healthcare.md

"energy","water","grid","transport infrastructure","railway","aviation","digital infrastructure" -> sector_critical_infra.md

"student","exam","grading","admission","proctoring","educational assessment","school","university" -> sector_education.md

"recruit","hiring","CV","candidate","performance review","termination","workforce","gig work" -> sector_employment_hr.md

"credit score","loan","insurance","fraud detection","financial risk","underwriting","mortgage" -> sector_finance_credit.md

"law enforcement","police","biometric","suspect","criminal","surveillance","remote identification" -> sector_law_enforcement.md (+ verify core_1 loaded)

"asylum","visa","migration","border control","refugee","immigration" -> sector_migration_border.md

"court","judge","sentencing","legal aid","election","voting","democratic process" -> sector_justice_democracy.md

"benefits","welfare","public service","entitlement","social security","essential service" -> sector_public_services.md

COMPLIANCE POSTURE DEFAULTS

Assume EUAIA applies until exclusion is confirmed
Assume HIGH RISK until risk classifier confirms otherwise
Assume DEPLOYER role at minimum — add others as confirmed
Disclose AI identity to any human user before interaction begins
Never take irreversible action without human confirmation
Log everything — if uncertain whether to log, log

ROLE GUIDANCE — HUMAN-FACING LAYER

11_ROLE_GUIDANCE/ contains guidance for HUMANS in each role — plain language explanations, decision trees, and actionable checklists. Load when a person needs to understand what the Act means for THEM, not just for the agent.

Roles covered: 01_guidance_provider.md — developers, engineers, product teams 02_guidance_deployer.md — business teams using third-party AI 03_guidance_compliance_lead.md — compliance officers, DPOs, legal, risk 04_guidance_management.md — CEOs, board, senior leadership 05_guidance_importer_distributor.md — supply chain, EU representatives

WHAT THIS FOLDER DOES NOT COVER

GDPR / Regulation (EU) 2016/679 — parallel, both apply NIS2 / Directive (EU) 2022/2555 — referenced in sector/hr_7 DORA / Regulation (EU) 2022/2554 — referenced in sector_finance MDR / Regulation (EU) 2017/745 — referenced in sector_healthcare Product Liability Directive (EU) 2024/2853 — separate folder needed National transposition measures — add per jurisdiction

UPDATE TRIGGERS FOR THIS FILE

DATE_TRIGGER: quarterly APPLICABILITY_TRIGGER: 2026-08-02 (full Act applies — mandatory sweep) APPLICABILITY_TRIGGER: 2027-08-02 (high-risk embedded products) EVENT_TRIGGER: any EU AI Office publication affecting Art. 1-4 or scope

When REVIEW_PENDING: agents continue under last APPROVED version. New Tier B skills not added until HITL approval.

REFERENCES

Art. 1, Art. 2, Art. 3, Art. 4, Art. 113 Source: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689