screen-trademark
について
このスキルは、複数の商標データベースを検索し、アバクロンビー・スペクトラムや混同の可能性テストなどの法的枠組みを適用することで、出願前の提案商標について、抵触と識別力の審査を行います。リスクマトリクスを含む抵触レポートを生成し、開発者が新たなブランド、ロゴ、スローガンを審査することを支援します。これは特に商標審査のために設計されたものであり、特許の先行技術調査とは異なるデータベースと法的手法を使用します。
クイックインストール
Claude Code
推奨npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/screen-trademarkこのコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします
ドキュメント
Screen Trademark
Screen proposed mark → conflicts + assess distinctiveness pre-filing. Search registries, eval Abercrombie, analyze confusion w/ priors, produce report w/ risk ratings.
Use When
- Pre-adopt new brand|product|service mark
- Rebrand|expand into new goods/services class
- Pre-file TM application (national|EU|intl)
- Eval acquisition w/ TM portfolios
- Pre-launch product new geo w/ existing brand
- C&D received → assess exposure
In
- Required: Proposed mark (word|figurative|both)
- Required: Goods/services covered (plain desc)
- Required: Geo scope (US|EU|specific|global)
- Optional: Nice classification classes if known
- Optional: First-use date (US common law priority)
- Optional: Known competing marks
- Optional: Word|figurative|composite
Do
Step 1: Define Mark + Goods/Services
Establish exact what + which classes.
- Record mark precisely:
- Word: text as appears (case for figurative)
- Figurative: visual elements, colors, stylization
- Composite: both
- Describe goods/services plain
- Nice classification:
- TMclass (https://tmclass.tmdn.org/)
- Search keyword → correct class + acceptable terms
- Most marks 1-3 classes; ID all relevant
- Adjacent (Class 9 software vs Class 42 SaaS)
- Geo scope:
- US (USPTO), EU (EUIPO), intl (WIPO Madrid), national offices
- US = first-to-use; EU = first-to-file
→ Record of mark, goods/services, Nice classes, jurisdictions. Defines search scope.
If err: Nice ambiguous → err on more classes. Broader screen safer than missing adjacent conflict.
Step 2: Search TM Databases
Identical + similar across registries.
- Identical first:
- TMview (https://www.tmdn.org/tmview/): EU + national
- WIPO Global Brand DB (https://branddb.wipo.int/): intl regs
- USPTO TESS / Trademark Center (https://tsdr.uspto.gov/): US
- National: DPMAregister (DE), UKIPO (UK), CIPO (CA)
- Similar — expand:
- Phonetic ("Kool" vs "Cool", "Lyft" vs "Lift")
- Visual ("Adidaz" vs "Adidas")
- Transliterations + translations
- Common prefix|suffix added|removed
- Plurals, possessives, abbrev
- Filter:
- Status: live|registered + pending (ignore dead|cancelled)
- Goods/services: same|related Nice
- Geo: target jurisdictions
- Per potential conflict record:
- Mark text + reg/app number
- Owner + jurisdiction
- Nice classes + goods/services
- Status (registered, pending, opposed) + dates
- Identical|similar (and how: phonetic|visual|conceptual)
→ List of potential conflicts from ≥2 DBs, identical+similar in relevant classes+jurisdictions. Each w/ enough detail for Step 4.
If err: DB unavailable → note gap, proceed w/ available. Common word → large result; prioritize same|closely related Nice before expanding.
Step 3: Distinctiveness
Where on Abercrombie spectrum.
- Abercrombie (weak→strong):
- Generic: Common name ("Computer Software" for software). Unregistrable
- Descriptive: Directly describes ("Quick Print"). Registrable only w/ secondary meaning
- Suggestive: Suggests but needs imagination ("Netflix"). Inherently distinctive
- Arbitrary: Real word in unrelated ctx ("Apple"). Strong inherent
- Fanciful: Coined w/ no prior meaning ("Xerox", "Kodak"). Strongest
- Secondary meaning if descriptive:
- Duration + extent of use
- Ad spend + consumer exposure
- Surveys|declarations
- Media + unsolicited recognition
- Genericide check:
- Was once distinctive but now common? ("escalator", "aspirin" US)
- Document w/ reasoning
→ Clear classification w/ rationale. Descriptive → assess if secondary meaning achievable. Suggestive|arbitrary|fanciful proceed confident.
If err: generic-descriptive border = significant risk. Recommend modify → push toward suggestive (add twist, combine unrelated) or prepare secondary meaning evidence strategy.
Step 4: Likelihood of Confusion
Eval if proposed likely confused w/ priors found Step 2.
- Per prior, assess DuPont (US) | EUIPO relative grounds:
- Mark similarity:
- Visual: side-by-side, letters, length, structure
- Phonetic: pronunciation, syllables, stress, vowels
- Conceptual: meaning, connotation, commercial impression
- Goods/services similarity:
- Same Nice = strong indicator not conclusive
- Related goods diff classes can conflict
- Trade channels + typical buyers
- Prior strength:
- Famous → broader (dilution doctrine)
- Weak|descriptive → narrower
- Market presence, ad spend, recognition
- Actual confusion evidence:
- Customer complaints, misdirected comms
- Social media confusion
- Prior opposition|cancellation
- Mark similarity:
- Weigh holistically:
- No single factor dispositive; balancing
- Strong mark sim can offset weak goods sim
- Famous tips toward confusion easier
- Rate per conflict:
- Blocking: Near-identical same goods, strong prior
- High: Similar same|related goods, or identical related
- Mod: Similar related, or identical distant
- Low: Weak similarity, distant goods, or weak prior
→ Rated conflict list w/ analysis. Most serious (blocking|high) w/ specific reasoning.
If err: borderline → conservative (higher risk). Safer to flag manageable than miss blocker.
Step 5: Common Law Rights
Unregistered rights not in DB searches.
- Search prior unreg use:
- Business name registries + state|provincial DBs
- Domain (WHOIS)
- Social handles + business profiles
- Industry directories + trade pubs
- Google + general web for commercial use
- Jurisdictional rules:
- US: First-to-use → prior commercial use creates rights w/o reg
- EU: First-to-file → reg priority but prior use limited defenses
- UK: Passing off doctrine protects unreg w/ goodwill
- Scope assessment:
- Geo reach of prior user's market
- Duration + consistency
- Goodwill built?
- Document + impact on overall risk
→ Supplementary unreg use list could create conflicts not visible in registry. Especially important US.
If err: common law overwhelming (common word) → focus same industry|category. Common law typically narrow scope — local bakery "Sunrise" doesn't block software "Sunrise."
Step 6: Goods/Services Overlap
Competitive proximity detail.
- Compare Nice vs each prior:
- Same class: presumptive overlap (not auto — classes broad)
- Adjacent: complementary or competitive?
- Distant: typically safe unless prior famous
- Trade channels:
- Sold via same retailers|platforms?
- Same demographic?
- Buyer assume common source?
- Expansion likelihood:
- Prior likely expand into proposed?
- "Zone of natural expansion" (US)
- Document w/ reasoning
→ Clear goods/services proximity per conflict, strengthening|weakening Step 4 ratings.
If err: relationship unclear (novel categories, convergent industries) → reasonable consumer test: typical buyer seeing both assume same source?
Step 7: Conflict Report
Compile all → structured actionable report.
- TM Conflict Report sections:
- Exec summary: proposed mark, key findings, overall risk
- Mark + scope: desc, Nice, jurisdictions
- Distinctiveness: Abercrombie, registration implications
- Conflict matrix: all conflicts + risk
Conflict Risk Matrix:
+----+-------------------+----------+---------+-------+---------+
| # | Prior Mark | Classes | Juris. | Type | Risk |
+----+-------------------+----------+---------+-------+---------+
| 1 | ACMESOFT | 9, 42 | US, EU | Ident | BLOCK |
| 2 | ACME SOLUTIONS | 42 | US | Sim | HIGH |
| 3 | ACMEX | 35 | EU | Phon | MOD |
| 4 | ACM | 16 | US | Vis | LOW |
+----+-------------------+----------+---------+-------+---------+
Risk: BLOCK = blocking | HIGH | MOD = moderate | LOW | CLEAR
Type: Ident = identical | Sim = similar | Phon = phonetic | Vis = visual
- Common law findings: relevant unreg use
- Goods/services analysis: per conflict overlap
- Recommendations — overall conclusion:
- Clear: No significant conflicts → proceed file
- Low: Minor conflicts unlikely block → proceed monitor
- Mod: Conflicts manageable → coexistence|modify|narrow goods
- High: Likely opposition|refusal → modify substantial or alt mark
- Blocking: Near-identical prior same goods → don't proceed w/o counsel
- Limitations + caveats:
- Screen ≠ legal opinion; consult counsel pre-file
- Common law may exist beyond DB
- Figurative similarity needs visual inspection (beyond text search)
→ Complete report w/ ratings, distinctiveness, recommendations. Enables go/no-go.
If err: inconclusive (mixed cross jurisdictions|classes) → present by jurisdiction; let decision-maker weigh business + legal. Qualified "proceed w/ caution" valid.
Check
- Mark + goods/services documented w/ Nice
- ≥2 TM DBs searched (TMview + USPTO TESS)
- Identical + similar searched (phon, vis, conceptual)
- Distinctiveness on Abercrombie w/ reasoning
- Confusion analyzed via DuPont|EUIPO
- Common law investigated (biz names, domains, web)
- Goods/services overlap per conflict
- Conflict matrix w/ ratings
- Overall recommendation (clear|low|mod|high|blocking)
- Limitations stated (screen vs legal opinion, DB gaps)
Traps
- Identical-only search: Misses dangerous — phonetic|visual sim trigger confusion. Always variants.
- Ignore related classes: Software (9) can conflict SaaS (42)|consulting (35). Nice = guidelines not walls.
- Skip common law: US — unreg w/ prior use trumps later federal reg. DB alone insufficient.
- Distinctiveness ≠ availability: Highly distinctive (fanciful) mark can conflict existing identical reg. Separate questions.
- Single-jurisdiction bias: Clear US may block EU. Always screen actual-use jurisdictions.
- Treat screen as legal opinion: Skill produces structured risk, not legal advice. Blocking|high warrant counsel review.
→
assess-ip-landscape— broader IP landscape mapping w/in full IP strategysearch-prior-art— patent-focused prior art (diff DBs + legal: novelty|obviousness vs confusion)file-trademark— filing post-screen (not yet avail)
GitHub リポジトリ
関連スキル
executing-plans
デザインexecuting-plansスキルは、完全な実装計画があり、それを管理されたバッチでレビューチェックポイントを設けながら実行する場合に使用します。このスキルは計画を読み込んで批判的にレビューした後、小さなバッチ(デフォルトは3タスク)でタスクを実行し、各バッチの間に進捗状況を報告してアーキテクトのレビューを受けます。これにより、品質管理チェックポイントが組み込まれた体系的な実装が保証されます。
requesting-code-review
デザインこのスキルは、コードレビュアーサブエージェントを起動し、処理を進める前に要件に対してコード変更を分析します。タスク完了後、主要な機能の実装後、またはmainブランチへのマージ前などに使用すべきです。このレビューは、現在の実装と元の計画を比較することで、問題を早期に発見するのに役立ちます。
connect-mcp-server
デザインこのスキルは、開発者がHTTP、stdio、またはSSEトランスポートを使用してMCPサーバーをClaude Codeに接続するための包括的なガイドを提供します。GitHub、Notion、カスタムAPIなどの外部サービスを統合するためのインストール、設定、認証、セキュリティについて解説しています。MCP統合のセットアップ、外部ツールの設定、またはClaudeのModel Context Protocolを扱う際にご利用ください。
web-cli-teleport
デザインこのスキルは、タスク分析に基づいて開発者がClaude Code WebとCLIインターフェースの選択を支援し、これらの環境間でのシームレスなセッションテレポーテーションを可能にします。Web、CLI、モバイル環境を切り替える際のセッション状態とコンテキストを管理することで、ワークフローを最適化します。様々な段階で異なるツールを必要とする複雑なプロジェクトにご活用ください。
