deciqAI Customs — HTS Classification (MECE Decomposition)
Industry front door for mece. Adds domain triggers, example, packs only. Parent Process unchanged. Not legal advice. Classification is fact-specific; use CBP rulings (CROSS) / binding rulings for certainty.
Activate when: classifying merchandise under the HTSUS; ambiguous/multi-function goods; "which heading applies?"; a product spanning multiple chapters. Do NOT activate when: an established binding ruling already governs the exact article.
Why this variant
The parent mece forces mutually-exclusive, collectively-exhaustive buckets. HTS classification via the General Rules of Interpretation (GRI) is literally a MECE walk: headings are meant to be exclusive; GRI 1→6 resolve overlaps so exactly one heading wins.
Domain inputs → parent's Process
- GRI 1: classify by heading terms + section/chapter notes (start here, exhaustively).
- If >1 heading fits (not mutually exclusive on its face): GRI 2 (incomplete/mixtures), GRI 3 (most specific → essential character → last in numerical order), GRI 4 (akin), GRI 5 (containers), GRI 6 (subheadings).
- The goal: a single, defensible heading with the reasoning recorded (reasonable care).
Worked example
A multi-tool with knife, screwdriver, LED light. → Not one obvious heading. Apply GRI 3(b) essential character; if indeterminate, GRI 3(c) last-in-order. Document the GRI path — that record is the reasonable-care defense in an audit.
Compliance anchors
- HTSUS + GRI; Section/Chapter Notes; 19 U.S.C. reasonable care; CBP CROSS rulings; binding-ruling option.
Packs
- Solo broker: GRI decision worksheet per SKU.
- Brokerage: ruling-backed classification library; binding-ruling triggers.
Red flags
- Jumping to a "close enough" heading without the GRI walk.
- Ignoring chapter notes that exclude a heading.
- No documented reasoning for a contestable classification.
Verification
- GRI 1 applied with section/chapter notes
- Overlaps resolved via GRI 2–6 in order
- Single defensible heading selected
- Reasoning recorded (reasonable care); binding ruling considered
Part of deciqAI Knowledge Skills. Core method: mece.
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/customs-hts-classification-mece · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/customs-hts-classification-mece.json