Ip Risk

IP Risk

Researches trademark/patent/design-patent risk for a candidate via legitimate public research, then records the result via hunt supplier-set --ip-status. Full vocabulary and human-origination rules live in references/supplier-ip.md; read it before using this Skill.

This Skill cannot, and must never claim to, provide legal clearance. That statement should be treated as load-bearing, not boilerplate — restate it explicitly to the human every time this Skill's findings are presented.

When to use this

For a promising candidate at COMPETITOR_ANALYSIS or later, before it reaches HUMAN_REVIEW — IP risk should be surfaced before a human is asked to make a final decision, not discovered after.

Step 1 — Check what's already recorded

hunt supplier-show <opportunity_id>

Review the existing ip_status/ip_notes before starting new research (time-budget-controller.md Rule 2).

Step 2 — What legitimate research looks like

• Public trademark databases — e.g. USPTO's TESS/trademark search, or equivalent public registries for the target marketplace. Search for the candidate's likely brand name/product name and closely related terms.
• Obvious existing-brand conflicts — is this product a close copy of an already-branded, actively-sold product from an identifiable brand? A clearly obvious, undisputed conflict (e.g. reusing a well-known brand name or an unmistakably copied proprietary design) is a real finding worth flagging distinctly from an ordinary "many sellers in this category" observation.
• Category-specific regulatory requirements — e.g. FDA requirements for products with food/skin contact, CPSIA for children's products, or similar category-specific compliance obligations that are publicly documented requirements, not opinions. Note what's required, not whether the candidate would pass — that determination is out of scope for this Skill.

All of the above is public-record and public-search research — reading public databases and product listings, not accessing anything restricted or automating around access controls.

Step 3 — What this Skill can conclude, and what it categorically cannot

Per references/supplier-ip.md SUP-01/IP-01, only these four literal values are valid for --ip-status:

UNKNOWN | PENDING_HUMAN_LEGAL_REVIEW | CLEARED_BY_HUMAN | CONFIRMED_BLOCKER

This Skill's own research, by itself, can only ever justify recording:

• PENDING_HUMAN_LEGAL_REVIEW — the normal outcome for essentially all research this Skill performs, whether findings look concerning or not. This is the honest status for "meaningful IP/compliance verification is incomplete," per CLAUDE.md's IP/COMPLIANCE section, and it is what a competent, careful web search can actually support.
• Flagging a genuinely obvious, undisputed conflict for human/lawyer confirmation of CONFIRMED_BLOCKER — e.g. an unmistakable trademark reuse or an unmistakably copied proprietary design. This Skill should surface such a finding prominently and recommend the human treat it as serious, but it must not call hunt supplier-set --ip-status CONFIRMED_BLOCKER itself. Only after a human (ideally with legal judgment, or at minimum deliberate human confirmation) reviews and agrees should that status be recorded.
• CLEARED_BY_HUMAN — this Skill never records this status itself either; by definition it requires a human's own review and clearance, which this Skill cannot substitute for.

CONFIRMED_BLOCKER is the only IP status that fires the hard gate (IP-01) — it must never be set from this Skill's own judgment alone, no matter how confident the research makes the finding feel.

hunt supplier-set <opportunity_id> --ip-status PENDING_HUMAN_LEGAL_REVIEW \
  [--ip-notes "<summary of what was searched and found>"]

Step 4 — Presenting findings

Always state plainly, in every findings summary:

• What was actually searched (which databases, which brand/product terms).
• What was found, including negative results ("no obvious trademark conflict found in TESS for [term]" — an absence-of-finding, not a clearance).
• That this research does not constitute legal clearance and a real IP-blocker finding needs a human, ideally a lawyer, to confirm CONFIRMED_BLOCKER before it can affect the candidate's standing.

What this Skill must never do

• Never claim legal clearance for a product, under any phrasing.
• Never call hunt supplier-set --ip-status CONFIRMED_BLOCKER or CLEARED_BY_HUMAN from its own judgment alone — both require explicit human confirmation first.
• Never treat "I didn't find an obvious conflict" as equivalent to "cleared" — that is exactly PENDING_HUMAN_LEGAL_REVIEW, not CLEARED_BY_HUMAN.

See references/supplier-ip.md for the full vocabulary and the rule-preservation table (SUP-01, IP-01 in particular).