Aulvem/epr-compliance

EPR + De Minimis Tariff Compliance Dataset (Cross-Border, EU/UK/US)

A structured, machine-readable dataset of packaging Extended Producer Responsibility (EPR) regimes and de minimis customs thresholds across EU, UK, and US, designed for cross-border e-commerce sellers and AI agents.

Overview

The cross-border compliance landscape changes constantly: EPR regimes expand scope, de minimis thresholds get suspended or reformed, EU PPWR phases in. Most existing online resources are human-readable narrative — almost none are usable by AI agents or typed applications without scraping. epr_compliance fills that gap with a typed schema, explicit current_state flags, and an append-only timeline per record that records what already happened AND what is scheduled.

Every record cites primary government sources (with reputable named secondary sources accepted for fast-moving transitions per SKILL.md §2.13), carries an explicit last_checked_at date, and a confidence rating.

Reference data, not legal advice. Consult a qualified compliance attorney, customs broker, or local regulatory authority before taking any business action.

What's in v0.1

Metric	Value
Records	14
EPR records	7 (DE / FR / IT / ES / NL / GB / US)
de_minimis_tariff records	7 (same geographies)
Timeline events	84
Confidence: high	11
Confidence: medium	3
Languages	English (master) + Japanese (view)
Last checked	2026-04-27
Schema version	0.1
License	CC-BY 4.0

The 7 geographies covered:

1. Germany (DE) — VerpackG / ZSVR / LUCID
2. France (FR) — AGEC / CITEO / Triman
3. Italy (IT) — D.Lgs 152/2006 / CONAI
4. Spain (ES) — Royal Decree 1055/2022 / Ecoembes
5. Netherlands (NL) — Packaging Management Decree 2014 / Verpact
6. United Kingdom (GB) — pEPR Regulations 2024 / PackUK
7. United States (US, CA representative) — SB 54 / CalRecycle / CAA

How to use

Python (datasets library)

from datasets import load_dataset

ds = load_dataset("Aulvem/epr-compliance")
print(ds["train"][0])
# {'id': 'epr_de_packaging', 'record_type': 'packaging_epr', 'country': 'DE', ...}

Direct JSON fetch (preserves nested structure)

The CSV is the default split (flattened union of both record types, with the timeline collapsed to a latest-event view), but the JSON file preserves the original structure including the full timeline array and the oneOf discrimination by record_type. Use the JSON if your code needs to filter by record_type or walk the complete event log.

import json, urllib.request

url = "https://huggingface.co/datasets/Aulvem/epr-compliance/resolve/main/epr_compliance_v0.1.json"
data = json.load(urllib.request.urlopen(url))

# Filter by record type
epr_records = [r for r in data["records"] if r["record_type"] == "packaging_epr"]
tariff_records = [r for r in data["records"] if r["record_type"] == "de_minimis_tariff"]

# Walk the timeline of one record
for ev in data["records"][0]["timeline"]:
    print(ev["date"], ev["event_type"], ev["description_en"])

Schema validation

import json, urllib.request, jsonschema  # pip install jsonschema

base = "https://huggingface.co/datasets/Aulvem/epr-compliance/resolve/main"
schema  = json.load(urllib.request.urlopen(f"{base}/schema.json"))
dataset = json.load(urllib.request.urlopen(f"{base}/epr_compliance_v0.1.json"))
jsonschema.validate(dataset, schema)   # raises on first failure

Schema

The full JSON Schema (Draft 2020-12) ships in this dataset as schema.json. Two record types share the same array, discriminated by record_type.

Common fields (both types)

Field	Type	Notes
id	string	epr_<cc>_packaging or tariff_<cc>_de_minimis
record_type	const	"packaging_epr" or "de_minimis_tariff"
country	string	ISO 3166-1 alpha-2 (uppercase)
country_name_en / country_name_ja	string	English master + Japanese view
notes_en / notes_ja	string	Material context and carve-outs
source_urls	array of URI	All authoritative sources cited
last_checked_at	string (ISO date)	Verification date
confidence	enum	high / medium / low / unknown
current_state	object	{ status, as_of }; status enum active / suspended / scheduled_for_removal / removed / unknown
timeline	array of object	Append-only event log. Substantive events sorted descending; initial_verification foundation marker pinned to array tail

packaging_epr specific

Field	Type	Notes
regulation_name_en	string	Original-language name (e.g. VerpackG, Triman); not translated
regulation_name_ja	string	Japanese gloss only — original remains canonical
registration_required	enum	yes / no / conditional / unknown
registration_authority_en/ja	string	Registering body (e.g. ZSVR for DE)
registration_url	string (URI)	Single canonical entry-point URL
covered_categories	array of enum	paper, cardboard, plastic, glass, metal, wood, composite, etc.
reporting_frequency	enum	annual / biannual / quarterly / monthly / on_threshold / unknown
marketplace_liability	enum	Whether marketplaces (Amazon, Etsy) bear seller liability
penalties_en/ja	string	Verbatim penalty range with original currency

de_minimis_tariff specific

Field	Type	Notes
threshold_value	number | null	Numeric threshold
threshold_currency	string | null	ISO 4217 (always paired with threshold_value)
threshold_value_jpy_estimate	number | null	Approximate JPY equivalent (not authoritative)
applies_to_categories	array of string	Goods categories (default ["all_goods"])
vat_above_threshold	enum	Whether VAT / sales tax applies above threshold
duty_above_threshold	enum	Whether import duty applies above threshold
special_regimes	array of string	Voluntary regimes (IOSS, OSS, etc.)
reporting_burden_en/ja	string	Free-text declaration summary

The CSV at epr_compliance_v0.1.csv is a flattened 37-column view of both record types (record_type-specific fields appear empty for the other type; the timeline is collapsed to a latest-event view via latest_event_* and timeline_event_count columns). The complete timeline is preserved in JSON only. Both files are kept bit-identical with the canonical sources in the GitHub repository.

Important notice

Reference only. Always verify with official sources before taking compliance, customs, or tax actions.

Cross-border compliance rules change frequently and have direct legal and financial consequences. Confirm directly with qualified attorneys, customs brokers, or local regulatory authorities before making business decisions. The dataset maintainers accept no liability for outcomes derived from this data.

Update cadence

The dataset is automatically re-fetched and re-extracted weekly (Mondays 03:00 UTC) by a GitHub Actions workflow in the source repository. Each run produces a pull request against the canonical GitHub repository with a structured diff against the prior snapshot. Updates are merged into the canonical dataset only after human review, and the merged result is mirrored here on Hugging Face.

You can pin to a specific revision using the standard HF dataset versioning, e.g. load_dataset("Aulvem/epr-compliance", revision="<commit-sha>").

Roadmap

Version	Focus
v0.1 (current)	14 records, packaging EPR + de minimis tariff
v0.2	WEEE EPR, textile EPR; broader marketplace liability data
v0.3	Asia-Pacific (JP / KR / TW) and additional US states
v1.0	Public REST API for cross-border compliance queries
v2.0	MCP server for direct AI-agent integration

Schema changes are additive — v0.1-pinned consumers will not be broken by later versions.

Source repository

Canonical source, weekly extraction pipeline, schema definition, and contribution process all live on GitHub:

https://github.com/aulvem/epr_compliance

Issues, pull requests, and inaccuracy reports should go there. This Hugging Face dataset is a downstream mirror.

How to cite

If you use this dataset in research, products, or downstream tooling, please cite as:

epr_compliance contributors (2026). epr_compliance: AI-readable EPR + De Minimis Tariff Compliance Dataset (v0.1). Available at https://huggingface.co/datasets/Aulvem/epr-compliance (mirror) and https://github.com/aulvem/epr_compliance (canonical), licensed under CC-BY 4.0.

CC-BY 4.0 requires attribution; the citation above — or any substantively equivalent form that names the dataset, version, and license — satisfies that requirement.

Pro edition (commercial / audit-grade)

The free baseline above is the AI-readable structured layer. The Pro edition adds audit-grade metadata that compliance teams need:

• Extraction evidence — original source URLs, fetched timestamps, and snippets supporting each field
• Confidence rationale — documented decision factors usable directly in ISO / SOC 2 audit trails
• Change history — structured version log for tracking regulatory amendments over time
• Cross-references — links to related EU regulations and other Aulvem datasets
• Compliance officer notes — pitfalls, recommended actions, official contacts, time estimates

Three tiers:

Tier	Price	Use case	Get it
Snapshot	$19	Personal / single-user research	aulvem.gumroad.com/l/xqtzbb
Subscription	$79 / year	Internal corporate use + 52 weekly updates + 4 quarterly briefs	aulvem.gumroad.com/l/hnkxj
Commercial License	$499 / year	Redistribution / white-label / REST API rights + 1 annual customization	aulvem.gumroad.com/l/mhxxhf

The free edition (this dataset) remains permanently free under CC-BY 4.0 and continues to receive weekly updates. Pro is a strict superset; Pro purchase does not replace your access to the free edition.

License

Released under Creative Commons Attribution 4.0 International (CC-BY 4.0). You are free to share and adapt, including for commercial use, provided you give appropriate credit.

The underlying regulatory text remains the property of each government and regulator. This dataset extracts factual information (registration thresholds, fee structures, deadlines, penalty caps) under principles applicable to factual extraction; it does not redistribute the original legislative or regulatory prose. If you republish raw regulator-authored text obtained via the source URLs, observe each regulator's own terms.

Contact

Questions, bug reports, and corrections via GitHub issues:

https://github.com/aulvem/epr_compliance/issues

Please cite the official source page when reporting an inaccuracy.