Domain name

Domain Name

Naming stalls more launches than code does. This skill turns it into a one-session decision: generate candidates against real criteria, check what's actually available (not guessed — checked), and pick.

Context

Read product-context.md if present. Otherwise ask for:

1. What the product does (one sentence) and 2–3 words users associate with it
2. Whether the product already has a working name, or naming is open too
3. Budget tolerance: standard registration (~$10–20/yr) only, or premium domains in play

Process

1. Generate candidates

Produce 12–20 candidates across these patterns (aim for 3–4 per pattern, skip patterns that don't fit):

• Literal: what it does (invoicefast, launchvideo)
• Compound: two short words (shipfast, solomax)
• Modified real word: drop/swap letters, add-a-suffix (shiply, launchr — use sparingly, dated fast)
• Invented but pronounceable: 2–3 syllables, spellable after hearing it once
• Verb-able: could a user say "just [name] it"?

Score each candidate against the say-spell-remember test:

• Say it aloud — does it survive "what's the site called?" in conversation?
• Spell it from hearing — no ambiguous letters (no "is that -ly or -ley?")
• Remember it tomorrow — distinct from incumbents in the space

2. TLD strategy for indie products

Pick target TLDs before checking. Defaults for a vibe-coded product:

• .com — still the trust default; check it first even if you won't buy it (a taken .com with an active competitor product is a rename signal)
• .app / .dev — good for products/tools; both force HTTPS (preloaded HSTS), fine
• .io / .co — accepted for SaaS; .io renewals are pricier
• .ai — only if AI is the product, not just the implementation; expensive (~$70–100/yr)

Avoid for a first product: hyphens, numbers, obscure TLDs users have never seen (.xyz is borderline), and any name where you'd have to say "no, spelled…"

3. Check availability (plan → validate → decide)

Run the bundled checker on the full candidate list — never assume availability from intuition; short names you "invent" are taken more often than not:

node scripts/check_domains.mjs taskflow shipfast getlaunch --tlds com,app,io

• Bare names expand across --tlds (default: com, app, io, dev, ai, co); full domains (taskflow.com) are checked as-is; --json for machine-readable output.
• Availability comes from RDAP (authoritative registry data, no key needed). 404 = available.
• If CLOUDFLARE_API_TOKEN and CLOUDFLARE_ACCOUNT_ID are set, the script also returns live registration/renewal pricing from the Cloudflare Registrar API and flags premium-tier names. Live pricing is the source of truth — RDAP-available names can still be premium-priced at the registry.

Present results as a shortlist table: only candidates with at least one good TLD available, with price where known.

4. Collision checks (before buying)

For the top 2–3 candidates:

1. Trademark: search the name at tmsearch.uspto.gov (and EUIPO if selling in the EU) in the software/SaaS classes. An exact live mark in your category kills the candidate — a lawyer letter after launch costs more than a rename now.
2. Active competitor: Google "[name]" software and "[name]" app. A dead domain squatter is fine; an active product with users is not.
3. Social handles: check X/Twitter, GitHub, and the one platform your audience lives on. Exact match not required — get[name], [name]app, [name]hq are accepted conventions.

5. Decide and register

Recommend one name with a one-paragraph rationale (memorability, availability, collision-free, price). Tell the user to register it today — good names available at standard price disappear, and re-running this process next week costs more than $12.

Gotchas

• RDAP 404 means unregistered, not necessarily buyable at standard price — premium names show as available but cost 10–100×. The Cloudflare pricing path exposes this via the tier field.
• Some ccTLDs (notably .io and .ai) have no public RDAP; the script falls back to a DNS check for these. taken via DNS is reliable; likely available (NXDOMAIN) still needs registrar confirmation — never tell the user a non-RDAP domain is definitively available.
• Don't batch more than ~30 domains per run out of courtesy to rdap.org; the script already limits concurrency to 4.

Output

1. Candidate table — all generated names with say-spell-remember scores
2. Availability results — the script's output, filtered to viable candidates with pricing where available
3. Collision check summary — trademark / competitor / handle status for the top 2–3
4. Recommendation — one name, one paragraph of rationale, and the registrar action to take today

---

Domain secured? When you're ready to launch, SoloMax turns your product URL into a launch-video pack — hooks, captions, and posting copy included.