An open-source compliance copilot for the EU Forced Labour Regulation (Regulation (EU) 2024/3015, "FLR").
Naming note: FLR here refers to the EU Forced Labour Regulation — not the FLR fisheries modelling framework and not the Free Law Reporter.
From 14 December 2027, products made wholly or partly with forced labour may not be placed on, made available in, or exported from the EU market — regardless of sector, company size, or where in the supply chain the forced labour occurred. Investigations are led by the European Commission (for cases outside the EU) or by national competent authorities, and operators may be asked to evidence their supply-chain due diligence within tight deadlines (30 working days at the preliminary stage).
Small and mid-sized exporters are caught in the middle: EU customers send them questionnaires, consultancy is expensive, and there has been no open-source starting point.
flr-agent aims to be that starting point — an open-source assistant that helps operators:
- Screen product categories and sourcing geographies against publicly documented forced-labour risk factors
- Understand the regulation in plain language, grounded in official texts
- Prepare due-diligence questionnaires and evidence for information requests
Important
flr-agent is not legal advice and is not affiliated with the European Commission. See DISCLAIMER.md.
This repository is deliberately small but runnable. What works today:
- ✅ Rule-based risk screening CLI (
country × sector→ auditable score + documented findings) - ✅ Curated list of public data sources (docs/data-sources.md)
- ✅ Plain-language FLR 101 (docs/flr-101.md)
- 🚧 LLM agent layer (LangGraph), RAG over the regulation & guidelines, questionnaire generation — see the roadmap
The bundled dataset is a small illustrative subset of publicly documented findings. You must extend and verify it against the official sources before relying on it.
Requires Python 3.10+.
pip install -e .
flr-agent screen --country CN --sector cottonOr without installing:
python -m flr_agent screen --country CN --sector cottonExample output:
FLR risk screening — flr-agent v0.1 (illustrative seed data)
──────────────────────────────────────────────────────────────
Country : CN
Sector : cotton
Score : 90/100 → HIGH
Documented findings for this country-sector pair:
• [HIGH] cotton / CN
State-imposed forced labour documented in Xinjiang cotton harvesting.
Sources: ILO findings (2024), UN OHCHR assessment (2022), U.S. UFLPA
Risk indicators typically associated with this sector:
- state-imposed forced labour
...
Recommended due-diligence actions:
1. Map the supply chain down to raw-material origin
...
Note: absence of a documented finding is NOT evidence of absence.
This tool is not legal advice.
Other commands:
flr-agent screen --country BD --sector garments --json # machine-readable output
flr-agent list-sectors # sectors in the bundled dataset| Doc | What it covers |
|---|---|
| docs/flr-101.md | The regulation explained: scope, timeline, enforcement process |
| docs/data-sources.md | Public data sources for forced-labour risk screening |
| docs/glossary.md | FLR terminology (economic operator, SIFL, substantiated concern, …) |
| docs/roadmap.md | Versioned roadmap, aligned with the regulation's own timeline |
The highest-value contribution at this stage is data: documented country-sector findings with verifiable public sources, and new sector profiles. See CONTRIBUTING.md.