/brand-name-checker

name: brand-name-checker
description: "Check product and brand names for conflicts across trademarks, domains, social handles, and package registries. Returns a risk level and Proceed/Modify/Abandon verdict. Don't use for name brainstorming, logo design, or trademark filings."
license: MIT
effort: max
metadata:
  version: 1.5.0
  author: "Luong NGUYEN <luongnv89@gmail.com>"

Brand Name Checker

Check product and brand names for conflicts across trademarks, domains, social media, and package registries (npm, PyPI, Homebrew, apt).

When to Use

Use before adopting a product or brand name. To save context budget, queries, edge cases, and templates live in `references/*.md`.

Subagent Architecture

Workers in `agents/` each return JSON: **social-checker** (6 platforms), **registry-checker** (npm, PyPI, Homebrew, apt), **domain-checker** (TLDs), **trademark-checker** (WIPO, EUIPO, INPI). The **synthesizer** applies the Step 5 risk policy.

• **Flow**: Social → (if no exact handle is taken) → parallel {Registry, Domain, Trademark} → Synthesizer.
• **Early-Exit Rule**: if social-checker finds an exact handle taken on any of the 6 platforms, skip Steps 2-4 and go straight to the synthesizer with an "Abandon" verdict.
• **Without subagents**: run Steps 1-4 inline from `references/source-checks.md`, in the same order.

Environment Check

1. Before Step 1, confirm the WebSearch and WebFetch tools are available. 2. Run one test query. 3. If either check fails, stop and return the `BLOCKED` report from *Output Format* with the error.

Repo Sync Before Edits (mandatory)

This skill writes no files by default. Before creating, updating, or deleting files in a repository, run the stash-first sync in `references/repo-sync.md`.

Input

Read the name from `$ARGUMENTS`; if empty, ask the user. If `prd.md` exists, read it for industry, target registries, and target domains.

Analysis Protocol

Queries, check URLs, and status rules for Steps 1-4: `references/source-checks.md`. Record every result with its source.

Step 1: Social Media Check (First Priority)

Search X/Twitter, Instagram, GitHub, LinkedIn, TikTok, and Discord. If an exact handle is taken (Early-Exit Rule), return `NEGATIVE: Exact social handle taken (@platform)` and skip to Step 6.

Step 2: Package Registry Check (if Step 1 clear)

Check npm, PyPI, Homebrew, and apt as Available, Taken (owner, last publish date), Similar, or Unknown. If target registry intent is missing or unverifiable, record it as unknown rather than treating every registry as clear.

Step 3: Domain Check (if Step 1 clear)

Check `.com` first, then `.io`, `.app`, `.co`, `.eu`, `.fr`:

• Available: an authoritative availability result confirms the domain is unregistered
• Parked: Domain exists but is for-sale/parking
• Active: In use (flag if same industry)
• Unknown: the source is unavailable or only shows that no active site was found; do not infer availability

Step 4: Trademark Check (if Step 1 clear)

Search WIPO, EUIPO, and INPI in Nice Classes 9, 35, 42; record each mark as live or expired.

Step 5: Risk Assessment

This section is the single authority for qualitative risk policy. The synthesizer must read and follow this section rather than inventing a second matrix. Classify each source result as confirmed, clear, or unknown/unverifiable. Upstream worker labels and severity fields are evidence only; re-check their underlying facts and do not blindly trust numeric severity.

Qualitative risk matrix

Risk precedence is **High > Moderate > Low**. Select the highest matching trigger; never average findings. "Multiple handles" means **two or more** confirmed non-exact handle collisions on distinct platforms.

| Risk level | Highest matching trigger | |------------|--------------------------| | **High** | Exact social handle collision on any of the 6 platforms (the Early-Exit Rule); two or more confirmed non-exact social handle collisions; a confirmed collision on **any target registry**; an active `.com` in the same industry; an active trademark conflict in Nice Classes 9, 35, 42; a well-known-brand/typosquat conflict; or the existing elevated trademark-collision risk for a very short name. | | **Moderate** | No High trigger, but one confirmed non-exact social handle collision; a different-industry active `.com`; a similar trademark; a confirmed collision on a non-target registry; or any unknown/unverifiable check or target intent. | | **Low** | Every required check and target intent is explicitly verified, with no High or Moderate trigger and no unknown result; `.com` is verified available or parked. |

Unknown policy: if checks or target intent are missing or unverifiable and there is no known High finding, use **provisional Moderate + Modify** pending verification. Unknown is not a confirmed collision and never Low clearance. An existing High finding remains High. No averaging or mitigation reduces High. Checks skipped by the Early-Exit Rule remain skipped, not cleared.

The exact-handle Early-Exit Rule is unchanged: an exact handle collision on any of the 6 platforms immediately returns an Abandon recommendation and skips the remaining checks. Report those checks as skipped, not clear.

Recommendation semantics

• **Proceed**: only for a fully verified Low result; state the evidence and limitations, and never represent unknown checks as clearance.
• **Modify**: for Moderate, including provisional Moderate; suggest variants that address confirmed conflicts and identify pending verification.
• **Abandon**: for High; mitigation or available alternatives cannot lower the High result.
• Any alternative that was not checked through the applicable sources must be labeled **unverified**; never invent availability or a Low assessment.

Step 6: Recommendation

Follow the recommendation semantics above and the registration order in Final Action.

Output Format

Return this compact text report, the only format: it fi