| name | name-product |
| description | Generate and screen a name for a new Minsky primitive — operand, sub-product, microsite, campaign, or marketing surface — using the Lexicon Branding methodology (three-phase identify/invent/implement, sound-symbolism, compound multiplier, processing fluency, polarization test, the diamond exercise, the synchronicity drill) with a mandatory prior-use / trademark-collision screen before any candidate advances. The synthesis dual of analyze-adjacent-product: that skill reads an existing name semiotically, this one generates a new name. Composes with minsky-brand (locked register constraint) and marketing-site-design (surface application). Use when naming a new operand, sub-product, or marketing surface, or evaluating an adjacent product's name. |
| user-invocable | true |
name-product — generate and screen a name using the Lexicon framework
You are naming a new Minsky primitive — a new operand for the locked vocabulary, a sub-product, a microsite, a campaign, or a marketing surface — or you are evaluating a naming choice (ours or an adjacent product's). This skill is the synthesis dual of /analyze-adjacent-product: that skill reads an existing name to extract what it constructs semiotically; this skill generates a new name and screens it before it ships.
If you are reaching for the first descriptive label that "says what it does" (task-runner, review-manager), stop. Descriptive names forfeit both the cumulative and the asymmetric advantage (see minsky-brand §7.1, principle 1). The method below is the resource-constrained single-agent dual of Lexicon's full process — enough rigor to avoid the obvious failure modes without a 1,500-candidate agency engagement.
Rename-neutral scope. This skill encodes the naming methodology. It does NOT decide whether Minsky-the-head-noun should be renamed — that is the active mt#2046 workstream's call (umbrella: "Brand-name exploration — should Minsky be renamed?"; position paper Draft at Notion 371937f03cb481e1983ded72834510ec; memory 2021067b). The trademark-collision screen below applies to every candidate including the current name, but the keep/rename verdict is mt#2046's, not this skill's.
When to invoke
- New operand for the locked vocabulary (
minsky-brand §7) — a term that compounds with the referent (a new Minsky ).
- New sub-product that needs its own name.
- New marketing surface — microsite, campaign, launch handle (compose with
/marketing-site-design §7a, the surface cut).
- Evaluating an adjacent product's name — when an
analyze-adjacent-product read surfaces a naming question you want to score against the framework.
- Sanity-checking a name already chosen — run the polarization / compound / code-coherence / trademark checks retroactively; a name that fails them is a candidate for change.
1. The framework — three phases
Lexicon's process is identify → invent → implement. The single-agent form below compresses it but keeps the load-bearing moves.
Phase 1 — Identify
- Ask the bidirectional-behavior question (
minsky-brand §7.1, principle 5): "How should the marketplace behave toward this thing, and how should it behave toward the marketplace?" This replaces the generic "what are its values / positioning / mission" framing and converges with Minsky's principal-substrate axis (mt#1931).
- Run the diamond exercise (§3) to surface the message that becomes the name.
- Name the locked constraints. For a Minsky primitive, the register is already locked: Cyberbrain / Section 9 (
minsky-brand §2), the five-layer reference architecture (§4), and the existing operand set (Mesh / Cockpit / Section / Asks / Flock). The new name must cohere with that register, not invent a new one.
Phase 2 — Invent
- Generate in generative mode, not evaluative mode. Ask "what could we do with this name?" — never "is this name good?" Humans default to noticing-what's-wrong; that reflex kills candidates before they're developed. Generate a wide set first; evaluate later.
- Run the synchronicity drill (§4) to force adjacency to unrelated domains.
- Bias toward the construction rules in
minsky-brand §7.1:
- Sound-symbolism — pull noise-generating phonemes for signal. The per-phoneme table is the single source of truth in
/minsky-brand §7.1 (V / Z / X / B / S / K / M / L-N) — consult it there; this skill does not duplicate it. Note the locked-vocab gap analysis: the MINSKY head-noun carries zero V/Z/X high-intensity phonemes, so high-intensity operands can compensate.
- Compound multiplier — prefer a compound (Minsky ) where 1 + 1 = 3; resist the client-reflex pull toward a single short token.
- Make intangible tangible — for an abstract substrate, reach for a physical-operational noun (Mesh, Section, Cockpit, Flock), not a sophisticated coinage.
- Processing fluency — if coining, build from familiar morphemes (Vercel =
ver- + -cel), not abstract-from-engineering opacity (Codium).
Phase 3 — Implement
Screen the surviving candidates through the four tests (§2) AND the mandatory trademark-collision screen (§5). Only a candidate that passes all of them advances to a naming decision. For a marketing surface, derive the visual treatment afterward per marketing-site-design §7 Step 5.
2. The screening checks
Run each finalist through all four. Any failure is disqualifying or a flag, not a footnote.
- Polarization test. Does the name provoke argument on first hearing? Comfortable team consensus is the warning sign — it means the name is too safe (Andy Grove's Pentium principle). Argument is energy.
- Compound-multiplier check. Does the name multiply associations (1 + 1 = 3), or sit as a lone token where a compound would carry two semantic circles? For operands, prefer the compound with the locked referent.
- Processing-fluency check. If coined, are the morphemes individually familiar? If a reader can't decode the parts, the whole reads as opaque.
- Not-like-the-other-guys test. Run the competitor drill (§3): would a buyer say "they're not like the other guys" before knowing what the thing does? That predisposition-to-consider is the signal the name is working.
3. The diamond exercise
Four corners; iterate over days, not in an hour:
- Win — how do you define winning for this thing?
- Have to win — the assets already in hand (locked myth, product proof, the corpus).
- Need to win — the gaps; what's missing to win.
- Have to say to win — the message that closes the gap. This message is what becomes the name.
The competitor drill (used by the not-like-the-other-guys test): ask a buyer-archetype reader "We just got a new competitor and their name is X — what do you think about them?" This surfaces what the name does for the imagination and bypasses the evaluative reflex.
4. The synchronicity drill (single-agent rotation)
Lexicon runs three disguised teams in parallel: one knows the real product, a second thinks it's naming (say) a bicycle, a third thinks it's naming an industrial system. The forced adjacency to unrelated domains produces words the on-the-nose team never reaches (naming a sailboat → read hunting and flying magazines).
The single-agent manual form is a rotation: deliberately generate three candidate sets under three disguised framings, then pool them.
- True framing — name the thing as what it is (a Minsky cybernetic-substrate primitive).
- Disguised framing A — pretend you are naming something physical and unrelated (a piece of climbing gear, a navigation instrument, a hand tool). Generate names for that.
- Disguised framing B — pretend you are naming an industrial or institutional system (a rail network, a power grid, a survey instrument). Generate names for that.
Pool all three sets, then map the disguised candidates back onto the real thing. The cross-domain words are where the non-obvious, story-opening names live. Keep the framings genuinely separate while generating — collapsing them back to the true framing too early defeats the drill.
5. Prior-use / trademark-collision screen (MANDATORY)
No candidate advances — including the current name — until it passes a prior-use screen. A same-class registered mark is a disqualifier or a high-risk flag, not a footnote.
Motivating finding (2026-05-31). MINSKY is a LIVE, INCONTESTABLE registered mark in IC 042: dMetrics, Inc., U.S. Reg. 5855970 (Principal Register, standard-character, Sec. 8 & 15 accepted 2026-05-29, senior — first use 2015), the AI/NLP platform behind their ~$100M DoD contract. The original Lexicon analysis missed this collision entirely; the keep/rename implications are mt#2046's to weigh. This screen exists so the same gap never recurs at proposal time — every candidate gets screened before it advances. (NEEDS VERIFICATION — out-of-repo: these registration details come from USPTO/TSDR and memory 2021067b, last checked 2026-05-31. Trademark status is mutable; reconfirm via TSDR or counsel before relying on it for a naming decision.)
What to screen for. Any registered or pending mark in the relevant classes:
- IC 009 (downloadable / packaged software) and IC 042 (SaaS / cloud / AI software services) are the load-bearing classes for a Minsky primitive. A LIVE registered mark in either, for related goods/services, is a high-risk flag.
- "Same class" is necessary but not sufficient — the real test is likelihood-of-confusion + relatedness of goods. Treat a same-class LIVE registration as a stop-and-assess, not an automatic kill; surface it to the principal with the relatedness read.
Tooling (document the options; do NOT wire a tool as part of this skill's scope).
- Markbase — free, no-auth JSON API for fast risk-sizing:
GET https://api.markbase.co/search?q=<term>. Daily USPTO mirror; returns word mark, owner, IC class, status code, and serial/registration number per hit. Use first to size the risk. (Verified live 2026-05-31: HTTP 200, returns structured USPTO hits.)
- USPTO TSDR API — authoritative live-status confirmation by serial/registration number. Free key from
account.uspto.gov/api-manager; or use trademark counsel for a reliance-grade clearance. Use to confirm a Markbase hit's current status (LIVE / DEAD / pending; Sec. 8 & 15 incontestability).
- Trademark-search MCP server (
jordanburke/trademark-mcp-server, MIT, on npm as trademark-mcp-server) — a DEFERRED option to wire only if naming work becomes recurring enough to justify an MCP tool. Not part of this skill's scope.
Process. (a) Markbase-search each finalist; (b) for any same-class hit, confirm live status via TSDR or counsel; (c) record owner, class, status, and serial/registration number in the naming write-up; (d) surface same-class LIVE registrations to the principal with the relatedness read before the candidate advances. Owning a domain does NOT make a name defensible — screen owned candidates too (the 2026-05-31 pass struck Visor and Agentic on trademark grounds despite domain availability; see memory 2021067b).
6. Composes with
/minsky-brand §7.1 — locked-register constraint. The five principles, the sound-symbolism table (single source of truth — cited here, not duplicated), the compound-multiplier rule, processing-fluency, the three pre-add checks, and the naming-layer anti-patterns all live there. A new Minsky primitive must cohere with the Cyberbrain / Section 9 register (§2) and the five-layer reference architecture (§4). This skill is the end-to-end process; minsky-brand §7.1 is the constraint it operates inside./marketing-site-design §7a — surface application. When the thing being named is a marketing surface (microsite, campaign, launch handle), use §7a's marketing cut on top of this method, then derive the visual treatment per §7 Step 5./analyze-adjacent-product — analytical dual. That skill reads an existing name/brand semiotically (Peirce / Barthes / Oswald); this skill generates one. Feed an adjacent-product naming read into a naming decision of our own; the not-like-the-other-guys signal is the same signal read from the generating side./pz-voice — voice layer. A name lands inside prose; the Macx-register voice and no-go register inform how a candidate reads in a sentence. Compose when the name will appear in manifesto / About / marketing copy.
7. Trigger conditions
Invoke this skill when:
- Adding a new operand to the locked vocabulary (
minsky-brand §7).
- Naming a new sub-product.
- Naming a new marketing surface (with
marketing-site-design §7a).
- Evaluating an adjacent product's naming as input to an RFC, position paper, or landscape analysis (with
analyze-adjacent-product).
Cross-references
/minsky-brand §7.1 — locked-register constraint and the canonical sound-symbolism table (single source of truth)./marketing-site-design §7a — marketing-surface naming cut./analyze-adjacent-product — analytical dual (reads a name; this skill generates one)./pz-voice — voice layer.- Upstream sources:
feedback_confabulated_strategic_frame_to_justify_tactical_preference — discipline against post-hoc sound-symbolism rationale (the Pepsi/Arnell trap at the naming surface).- Originating tasks: mt#2045 (this skill), mt#2046 (rename umbrella), mt#2189 (trademark clearance, parked), mt#1929 (brand workshop umbrella), mt#1931 (principal-substrate axis), mt#1933 (minsky-brand), mt#1944 (analyze-adjacent-product), mt#1952 (pz-voice).
