Arness Spark Naming

Guide a product from nameless concept to validated brand name through a structured 4-step methodology, driven by the arn-spark-brand-strategist agent. Produces a naming brief (naming-brief.md) in the vision directory and a naming report (naming-report.md) in the reports directory.

Prerequisites

1. Read the project's CLAUDE.md for the ## Arness section.
2. Extract Vision directory and Reports directory paths. If no ## Arness section exists or Arness Spark fields are missing, inform the user: "Arness Spark is not configured for this project yet. Run /arn-brainstorming to get started — it will set everything up automatically." Do not proceed without it.
3. Create directories if they do not exist.

Step 0: Context Gathering

Product context

Check for <vision-dir>/product-concept.md:

If found: Read and extract: vision statement, value proposition, target audience, product pillars, competitive landscape. Summarize the extracted context to the user: "Found your product concept. I'll use this as the foundation for naming."

If not found:

Ask (using AskUserQuestion):

"No product concept found. How should I learn about your product?"

1. Describe your product — Provide a description in the next message
2. Point me to a file — Specify a file path containing product information
3. Explore current project — I'll read README, package.json, and code to infer what the product does

If option 3: invoke arn-spark-brand-strategist in brand-dna mode with instructions to explore the project and summarize the product context.

Target market

Ask (using AskUserQuestion):

"What is the primary target market? This determines trademark databases and languages for linguistic screening."

1. United States — USPTO trademark search, English + Spanish linguistic check
2. European Union — EUIPO trademark search, English + French + German + Spanish + Italian
3. United Kingdom — IPO trademark search, English linguistic check
4. Global / Multiple regions — WIPO + major national databases, all major languages

Existing naming brief

Check for <vision-dir>/naming-brief.md:

If found:

Ask (using AskUserQuestion):

"A naming brief already exists. How would you like to proceed?"

1. Resume from where I left off — Continue from the first incomplete section
2. Start fresh — Preserve existing as naming-brief-previous.md and begin new

If resume: read the brief, detect which sections contain "-- Pending --" or are missing, and resume from the first incomplete step.

---

Step 1: Strategic Foundation (Brand DNA)

Invoke arn-spark-brand-strategist in brand-dna mode with:

• Product context (from product-concept.md or user input)
• Competitive landscape (from product concept, or agent will research via WebSearch)
• Target market

The agent returns: brand personality profile, audience vocabulary, competitor name landscape, 1-2 recommended naming categories with rationale.

Present the Brand DNA analysis to the user.

Ask (using AskUserQuestion):

"Proceed with the recommended naming categories?"

1. Yes, proceed with [recommended categories] (Recommended) — Use the strategist's recommendation (substitute actual category names from the Brand DNA analysis)
2. Choose different categories — Select naming categories manually

If option 2:

Ask (using AskUserQuestion, multiSelect: true):

"Select naming categories to explore (select all that apply):"

1. Descriptive — Names that describe what the product does (PayPal, Dropbox)
2. Evocative — Names that suggest a feeling or metaphor (Slack, Nike)
3. Invented / Abstract — Completely new words (Spotify, Google)
4. Lexical — Wordplay, puns, alliteration (Pinterest, Netflix)

Then prompt (free-text): "Any existing name ideas, words you love, or words you hate? These will seed the creative sprint. Type 'none' or 'skip' to continue without seeds."

Write initial <vision-dir>/naming-brief.md using the creative brief template:

Read ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/references/creative-brief-template.md

Populate: Context and Brand DNA sections. Mark remaining sections as "-- Pending --".

---

Step 2: Creative Sprint (The Big List)

Four generation rounds via arn-spark-brand-strategist in generation mode.

Round 1 — Seed harvest: Invoke agent with user's existing ideas and preferences as seeds. If no seeds, skip to Round 2.

Round 2 — Category sprints: Invoke agent once per selected category, requesting 50-80 candidates each. Pass the brand DNA context and dead directions. Present candidates to user after each category sprint for early feedback.

Round 3 — Mashup round: Invoke agent with all Round 1-2 candidates. Cross-pollinate fragments across categories. Target: 30-50 mashup candidates.

Round 4 — User collaboration checkpoint: Present the complete candidate list organized by category. This is a free-text conversation loop: the user marks favorites (star), flags directions to kill, and may suggest additional directions. Iterate until the user signals satisfaction.

Target: 200+ total candidates across all rounds.

Update naming-brief.md with Creative Sprint Results section (generation stats, candidates by category, starred favorites, dead directions).

---

Step 3: Qualitative Filter (Six Senses)

Load the scoring methodology:

Read ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/references/naming-methodology.md

Invoke arn-spark-brand-strategist in scoring mode with: full candidate list, user-starred favorites, dead directions.

Pass 1: Agent filters 200+ candidates to 30-40 by removing obvious duds (unpronounceable, too long, too similar, offensive, dead directions). User-starred names always survive Pass 1.

Pass 2: Agent scores each surviving candidate on the Six Senses (1-5 each):

1. Appearance — visual punch, letter count
2. Sound — euphony, Phone Test
3. Meaning — associations, connotations
4. Memorability — stickiness, alliteration
5. Function — verbability, typability, handle-fit
6. Scalability — expansion headroom

Present the scored table sorted by total score (6-30).

Prompt (free-text): "Select 5-8 finalists for due diligence. You can pick from the top of the table, or choose names you like regardless of score."

Update naming-brief.md with Qualitative Filter Results section.

---

Step 4: Cold Shower (Due Diligence)

4a — Domain Availability

Load the WHOIS/RDAP server reference:

Read ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/references/whois-server-map.md

How it works: Both scripts use RDAP (the modern IETF standard) as the primary lookup method, with port-43 WHOIS and system whois as built-in fallbacks. The fallback chain per domain is: RDAP → port-43 WHOIS → system whois → manual URL. The scripts handle this automatically — no manual fallback logic needed.

Environment detection (priority order):

1. Check Python: run python3 --version. If available, use ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/scripts/whois-check.py.
2. Check Node.js: run node --version. If available, use ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/scripts/whois-check.js.
3. Neither available: skip automated checking. Present manual fallback URLs (https://www.whois.com/whois/[domain]) for each domain.

Note: System whois is NOT required — both scripts have it as a built-in last-resort fallback (gracefully skipped on Windows where whois is not available). Python and Node.js scripts work identically across Linux, macOS, Windows, and WSL2.

Rate limit discovery: Before running queries, use WebSearch for "[RDAP or WHOIS server name]" rate limit for the primary TLD servers involved. Set delay_seconds to the most conservative discovered limit. Floor: 2 seconds. Default if unknown: 3 seconds.

Domain list construction: Each finalist name × TLDs. Start with global TLDs: .com (essential), .io, .co, .dev, .app, .ai. Then add country-specific TLDs based on target market (see ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/references/whois-server-map.md for the market-to-TLD mapping):

• US: add .us
• EU: add .eu, .de, .fr, .it, .es, .nl
• UK: add .co.uk
• Global: add relevant local TLDs based on user's primary markets

The scripts handle compound ccTLDs (.com.br, .co.uk, .com.au, etc.) and RDAP-only TLDs (.dev, .app, .land) automatically. Ask user if additional TLDs should be checked.

Execution:

echo '{"domains": ["name1.com", "name1.io", ...], "delay_seconds": N}' | python3 ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/scripts/whois-check.py

Parse JSON output. Each result includes a method field ("rdap", "whois", "system-whois", or "none") and a manual_url for any domain that couldn't be determined. If exit code 1 (RDAP rate limit circuit breaker): read partial results, report what was checked, offer manual URLs for unchecked domains.

4b — Trademark Screening

Load trademark database reference:

Read ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/references/trademark-databases.md

Tier 1 (automated): Use WebSearch for "[name]" trademark [industry] and "[name]" registered trademark for each finalist.

Tier 2 (guided): Based on the target market, generate direct search URLs from the trademark database reference. Present URLs as a clickable checklist for the user to verify manually.

Always include: "This is a preliminary screening. Consult a trademark attorney before committing to a brand name."

4c — Linguistic Screening

Invoke arn-spark-brand-strategist in linguistic-screening mode with: finalist names, target market, relevant languages (mapped from target market per the agent's language mapping).

The agent checks each name for: negative meanings, phonetic conflicts, cultural associations, and slang issues. Uses WebSearch for verification of uncertain findings.

4d — Social Media Handle Check

Use WebSearch for each finalist: "[name]" @[name] site:twitter.com OR site:x.com OR site:instagram.com OR site:github.com

Report availability: available, taken, or unknown for each platform.

---

Final Output

Update naming brief

Update <vision-dir>/naming-brief.md with:

• Due Diligence section (domain matrix, trademark results, linguistic results, social handles)
• Final Decision section (chosen name, rationale, runner-ups)

Write naming report

Load the report template:

Read ${CLAUDE_PLUGIN_ROOT}/skills/arn-spark-naming/references/naming-report-template.md

Write <reports-dir>/naming-report.md with all sections populated.

Update product concept (conditional)

Only if <vision-dir>/product-concept.md exists:

Present the proposed change: adding the brand name to the product concept's Vision section.

Ask (using AskUserQuestion):

"Update the product concept with the chosen brand name?"

1. Yes, update it — Add brand name to the Vision section of product-concept.md
2. No, keep as-is — Product concept remains unchanged

If Yes: read product-concept.md, add **Brand name:** [chosen name] to the Vision section, update the document title if it contains a placeholder. Write back.

Summary

Present to the user:

• Chosen brand name
• Naming brief location: <vision-dir>/naming-brief.md
• Naming report location: <reports-dir>/naming-report.md
• Product concept updated: yes / no
• Any due diligence gaps to address (unchecked domains, trademark databases to verify manually)
• Runner-up names for reference

---

Agent Invocation Map

Step	Agent	Mode	Purpose
1	arn-spark-brand-strategist	brand-dna	Analyze brand personality, audience, competitors, recommend categories
2 (each round)	arn-spark-brand-strategist	generation	Generate 50-80 candidates per category per round
3	arn-spark-brand-strategist	scoring	Filter and score candidates on Six Senses framework
4c	arn-spark-brand-strategist	linguistic-screening	Check finalists in target languages
4a	Direct (Bash)	scripts	Run WHOIS availability check
4b, 4d	Direct (WebSearch)	—	Trademark and social media screening

Error Handling

• WHOIS script fails: Read partial JSON results. Report checked vs. unchecked domains. Offer manual fallback: https://www.whois.com/whois/[domain]
• Neither Python nor Node.js available: Skip automated WHOIS entirely. Present manual check URLs.
• WebSearch fails during trademark screening: Note the gap. Present trademark database URLs for manual verification.
• User cancels mid-process: Save current progress to naming-brief.md (incomplete sections marked "-- Pending --"). Resumable via Step 0.
• Product concept update fails: Print the proposed change in conversation for manual application.
• Agent returns insufficient candidates: Retry with adjusted prompt (broader techniques, relaxed constraints). If still insufficient, present what was generated and ask user whether to continue or adjust direction.

Constraints

• Product-concept.md updates are NET-NEW information (brand name addition), not stress-test-driven modifications. This does not conflict with the concept-review exclusivity rule (from arn-spark-concept-review, which restricts stress-test-recommendation consolidation to that skill alone).
• User approval gate is MANDATORY before writing to product-concept.md.
• All reference/script paths use ${CLAUDE_PLUGIN_ROOT}.
• WHOIS queries use a circuit breaker — any error stops all remaining queries immediately to protect the user's IP.
• The naming brief overwrites on re-run (git provides history). The user is warned and offered resume/start-fresh in Step 0.