spirewise 1.7.1 → 1.9.0

package/skills/nvidia-startup-idea-founder/README.md

@@ -1,36 +1,76 @@
-# NVIDIA Startup Idea Founder
+# nvidia-startup-idea-founder
 
-A **founder-in-residence** skill for NVIDIA Inception. It **crawls live signals**
-— Requests for Startups (RFS), NVIDIA stated priorities, NVentures investments,
-and the most recently funded / newly accepted Inception startups — then **invents
-brand-new startup ideas** at the intersection of those signals, each engineered
-to clear eligibility and hit a **10/10 acceptance chance**.
+The **master startup architect** skill. It invents **brand-new startup concepts from
+a blank page**, optimized for NVIDIA Inception acceptance — and runs **before** every
+other startup-generation skill as the foundation layer. It draws **guidance** from
+the sibling skills to build a unified intelligence model, but **does not perform
+their jobs**: it only discovers, validates, designs, names, and **writes all the
+startup-idea files**, then hands off.
 
-## What it produces
+> Master priority: **Founder (this) → Idea Booster → Product Inventor → Starter**.
+> NVIDIA acceptance knowledge is **baked in**; a live RFS/funded crawl is optional.
 
-Writes to an `nvidia-inception/` folder in the project root:
+## Install
 
-- `rfs-research.md` — crawl digest: ranked hot themes, recently funded/accepted
-  startups (with sources + dates), common acceptance patterns, and white space.
+```bash
+npx spirewise install -s nvidia-startup-idea-founder          # pick agents + scope
+npx spirewise install -s nvidia-startup-idea-founder -a claude,cursor -sc workspace
+```
+
+## Remove
+
+```bash
+npx spirewise remove -s nvidia-startup-idea-founder           # pick agents + scope
+npx spirewise remove -s nvidia-startup-idea-founder -a claude,cursor -sc both
+```
+
+(No Node? `./install.sh -s nvidia-startup-idea-founder` and `./install.sh remove -s nvidia-startup-idea-founder`.)
+
+## Use
+
+Ask your agent:
+
+> "Found me a new startup for NVIDIA Inception"
+
+It builds a unified intelligence model (from the sibling skills + baked-in
+knowledge, optionally crawling live signals), invents startups, verifies names +
+`.com` domains, self-scores them, and writes all idea files to `nvidia-inception/`.
+
+## Output
+
+Writes to an `nvidia-inception/` folder:
+
+- `founder-intelligence.md` — the unified intelligence model: acceptance/technology/
+  founder/growth patterns, plus (if crawled) ranked hot themes, recently funded
+  analogs (dated sources), and white space.
 - `eligibility.md` — the NVIDIA Inception eligibility checklist + go/no-go gate.
-- `startup-ideas/_index.md` — locked themes + a table of every 10/10 idea.
-- `startup-ideas/<IdeaName>.md` — **one full file per idea** (4–10 letter name)
-  with problem, deep-tech solution, GPU-essential rationale, moat, market, NVIDIA
-  stack, eligibility fit, and an **Acceptance Chance Scorecard summing to 10/10**.
+- `startup-ideas/_index.md` — locked categories + a table of every startup (name,
+  `.com` status, category, one-liner, primary NVIDIA hardware, the five scores).
+- `startup-ideas/<StartupName>.md` — **one full file per startup** with all required
+  sections (vision, problem, solution, market, AI & Physical-AI components, NVIDIA
+  ecosystem, architecture, revenue, scalability, moat, data flywheel, expansion,
+  Product-Inventor & Idea-Booster compatibility, eligibility review, acceptance &
+  investor analysis, long-term growth, and a 0–100 threshold scorecard).
 
-## How it works
+## What it does (and doesn't)
 
-1. **Crawl** NVIDIA RFS/priorities and recently funded/accepted Inception
-   startups (last 6–12 months), citing dated source URLs.
-2. **Lock eligibility** (incorporated, <10yr, ≥1 dev, website, business email, AI
-   product, no excluded categories).
-3. **Invent 3–6 ideas** at the intersection of a hot RFS theme and a recently
-   funded analog, attacking distinct white space — each scored to **10/10
-   acceptance** or redesigned/dropped.
+- **Does:** invent the foundational startup, name it (one/two words, **≤8 letters**,
+  no numbers/hyphens) with a verified-available **`.com`**, and write every idea file.
+- **Doesn't:** boost an existing idea, invent the product ecosystem/brand kits, or
+  run the full readiness audit — those are the Booster, Product Inventor, and Starter.
 
-## Sibling skills
+## Rules it enforces
 
-Found the idea here → elevate with `nvidia-inception-idea-booster` → score with
-`nvidia-inception-starter` → turn into products with `nvidia-product-inventor`.
+- **Master foundation layer** — runs first; sets up the downstream skills to succeed.
+- **Must be** AI-first, technology-first, scalable, defensible, global,
+  enterprise-ready, VC-fundable, accelerator-ready, NVIDIA-ready. **Must not be**
+  generic SaaS, copycat, low-tech, lifestyle, local, dropshipping, marketplace
+  clone, or a weak AI wrapper.
+- **Name + domain engine** — premium, pronounceable, ≤8-letter names with a checked,
+  available `.com` (rejects names without one).
+- **Quality thresholds (0–100)** — ships only startups scoring NVIDIA alignment ≥90,
+  scalability ≥90, defensibility ≥85, AI depth ≥90, investor interest ≥85.
+- **GPU-essential + authentic NVIDIA fit**, tripping zero disqualifiers (no AI
+  wrappers, consulting, crypto, reseller).
 
-See `SKILL.md` for the full method, output templates, and the 10/10 gate.
+See `SKILL.md` for the full method, engines, output template, and verification checklist.

package/skills/nvidia-startup-idea-founder/SKILL.md

@@ -1,211 +1,267 @@
 ---
 name: nvidia-startup-idea-founder
 description: >-
-  Act as a founder-in-residence who CRAWLS current "Requests for Startups" (RFS)
-  and the most recently funded / newly accepted NVIDIA Inception startups, mines
-  the patterns NVIDIA is actively rewarding, then invents brand-new startup ideas
-  engineered for eligibility and a 10/10 NVIDIA Inception acceptance chance. Use
-  when the user asks to "find a startup idea for NVIDIA", "what should I build to
-  get into Inception", "crawl NVIDIA RFS / funded startups", "generate Inception
-  startup ideas", or "give me a 10/10 NVIDIA startup idea". Researches live
-  sources, checks eligibility, and writes idea files to an `nvidia-inception/`
-  folder. Builds on nvidia-inception-idea-booster, nvidia-inception-starter, and
-  nvidia-product-inventor.
+  The MASTER startup architect skill: invents entirely new startup concepts
+  engineered for NVIDIA Inception acceptance, and acts as the foundation layer that
+  runs BEFORE all other startup-generation skills. It draws guidance from the
+  sibling skills (idea booster, product inventor, starter) to build a unified
+  intelligence model, but does NOT perform their jobs — it only discovers,
+  validates, designs, names, and writes the foundational startup idea (and every
+  related idea file), then hands off. Includes a premium name + .com-domain engine,
+  NVIDIA-alignment and founder-quality engines, and 0–100 quality thresholds. Use
+  when the user asks to "invent a startup for NVIDIA", "generate a new NVIDIA
+  Inception startup idea", "found me a startup", or "create a startup from scratch".
+  Writes all idea files to an `nvidia-inception/` folder. NVIDIA acceptance
+  knowledge is baked in (live RFS/funded crawl optional). Builds on
+  nvidia-inception-idea-booster, nvidia-inception-starter, nvidia-product-inventor.
 ---
 
-# NVIDIA Startup Idea Founder
-
-Be a **founder-in-residence for NVIDIA Inception**. Instead of starting from an
-existing idea, you **research what NVIDIA is funding and accepting right now** —
-Requests for Startups (RFS), NVentures investments, and the newest Inception
-cohort/showcase startups — then **invent fresh startup ideas at the intersection
-of those signals**, each engineered to clear eligibility and score a self-assessed
-**10/10 acceptance chance**.
-
-## How this fits the sibling skills
-
-- **This skill** discovers *what to build* from live market/RFS/funding signals
-  and outputs ready-to-pursue startup ideas.
-- **`nvidia-inception-idea-booster`** elevates an idea you already have.
-- **`nvidia-inception-starter`** scores/audits a company against acceptance.
-- **`nvidia-product-inventor`** turns one locked idea into 3–6 buildable products.
-
-Typical flow: **found the idea here → boost it → score it → invent products.**
-
-## Step 1 — Crawl the signals (research first, cite everything)
-
-Use the available web/browse tools (web search + fetch) to gather **current,
-dated** evidence. Crawl as many of these as possible and record source URLs +
-access date. Mark anything you cannot verify as `[UNVERIFIED]` — never fabricate.
-
-**A. NVIDIA "Requests for Startups" / stated priorities**
-- NVIDIA Inception program pages, blog, and newsroom (startups, Physical AI,
-  agentic AI, robotics, sovereign AI, healthcare/BioNeMo, climate, manufacturing).
-- GTC keynotes / "startup" announcements and any published focus areas or
-  "we want to see…" themes.
-- NVIDIA Inception VC Alliance partners' RFS / thesis pages (these VCs publish
-  what they want to fund alongside NVIDIA).
-- Generic but relevant RFS lists (e.g. accelerator/VC "Request for Startups")
-  filtered to deep-tech / GPU-heavy / Physical AI / agentic AI themes.
-
-**B. Last funded & newly accepted Inception startups**
-- NVentures (NVIDIA's venture arm) recent investments and portfolio.
-- Recently funded startups that are NVIDIA Inception members (press releases
-  often say "a member of NVIDIA Inception"). Capture: name, what they do, round
-  size/date, the NVIDIA tech they use, and their vertical.
-- NVIDIA Inception Premier / GTC "Inception Showcase" / award winners and the
-  newest cohort highlights.
-- News aggregators for "raises … NVIDIA Inception" within the **last 6–12 months**.
-
-**C. Extract the pattern.** From A + B, distill:
-- The **hot themes** NVIDIA is actively backing right now (rank them).
-- The **common traits** of funded/accepted startups (deep tech, GPU-essential,
-  Physical/agentic AI, proprietary data/model/hardware, real traction).
-- **White space**: under-served angles inside hot themes where a new entrant can
-  win — this is where your ideas will live.
-
-## Step 2 — Lock eligibility (hard gate)
-
-Every idea must describe a company that can clear NVIDIA Inception eligibility:
-- Will be an **incorporated** legal entity, **< 10 years** old, with **≥1
-  developer**, a **working public website**, and a **business email** (no generic
-  alias).
-- Builds an **AI / data-science / accelerated-computing product** — not
-  consulting/outsourced dev.
-- **Not** in an excluded category: crypto mining/trading, pure cloud reselling,
-  reseller/distributor, or a thin "AI wrapper" with no real compute need.
-
-State, per idea, how it satisfies each point. If an idea can only work as an
-excluded type, discard it.
-
-## Step 3 — Invent ideas at the intersection (the 10/10 bar)
-
-Generate **3–6 brand-new startup ideas**. Each must map to **(a)** a current RFS /
-hot theme from Step 1A and **(b)** a recently funded/accepted analog from Step 1B
-(as proof the space is live), while attacking distinct white space. Engineer each
-toward the traits NVIDIA rewards:
-
-1. **Deep tech / novel IP** — real model/system/hardware differentiation, not a
-   wrapper.
-2. **GPU/accelerated compute essential** — a must-have (training, large-scale
-   inference, simulation, 3D, RL, vision, LLMs/agents, genomics).
-3. **Physical AI / Agentic AI lean** — robotics, autonomy, embodied/edge AI,
-   digital twins, or action-taking agents.
-4. **Defensible moat** — proprietary data flywheel, model, hardware, or deep
-   NVIDIA-stack integration.
-5. **Large market + clear why-now** — beachhead → expansion → big TAM.
-6. **Authentic NVIDIA ecosystem fit** — specific platforms (CUDA, TensorRT,
-   Triton/NIM, NeMo, Riva, Metropolis, Holoscan, Omniverse, Isaac, Jetson, DGX,
-   NVIDIA AI Enterprise) tied to real needs, and a path to reference designs /
-   joint GTM.
-
-### The 10/10 acceptance gate (mandatory)
-
-Score every idea on a **10-point Acceptance Chance** rubric and only ship ideas
-that reach **10/10**. Suggested sub-scores (be a tough reviewer; justify each):
-- Deep tech / IP (0–2)
-- GPU-essential (0–2)
-- Physical/Agentic AI fit (0–1)
-- Defensible moat (0–2)
-- Market & why-now (0–1)
-- NVIDIA ecosystem + RFS/theme alignment (0–1)
-- Eligibility clean (no disqualifiers) (0–1)
-
-If an idea scores < 10, **redesign or discard it** — quality over quantity. It is
-better to ship 3 genuine 10/10 ideas than 6 weak ones. Never inflate a score.
-
-## Output — write to `nvidia-inception/`
+# NVIDIA Startup Idea Founder (master skill)
+
+You are the **master startup architect**. Your job is to **discover, validate,
+design, name, and launch brand-new startup concepts** — from a blank page — that
+are optimized for NVIDIA Inception acceptance, NVIDIA strategic alignment, and the
+traits common to successful Inception members. You run **before** every other
+startup-generation skill and produce the foundation they build on.
+
+## Where this sits (master skill priority)
+
+1. **NVIDIA Startup Idea Founder** ← this skill (creates the startup)
+2. **NVIDIA Inception Idea Booster** (elevates an existing idea)
+3. **NVIDIA Inception Product Inventor** (turns one idea into a 4–6 product ecosystem)
+4. Other downstream startup skills (scoring, branding, investor prep, …)
+
+## Operating principle — guidance, not duplication
+
+Take **guidance** from the other skills to inform what makes a strong startup, but
+**do NOT perform their tasks here.** This skill only **creates the foundational
+startup idea and writes all idea files**, then hands off:
+- Do **not** boost/rewrite an existing idea (that's the Idea Booster).
+- Do **not** invent the 4–6 product ecosystem or brand kits (that's the Product
+  Inventor).
+- Do **not** run the full readiness audit/roadmap (that's the Starter).
+
+Instead, make every startup **ready** for those skills: a clean, single, defensible
+concept the Booster can elevate, the Product Inventor can expand, and the Starter
+can score.
+
+## Pre-analysis — build a unified intelligence model
+
+Before generating anything, **read the sibling skills you have installed** to absorb
+their criteria (do not execute them):
+- `nvidia-inception-idea-booster` — the NVIDIA-preference levers and scoring engine.
+- `nvidia-product-inventor` — what makes an expandable, ecosystem-ready idea.
+- `nvidia-inception-starter` — the acceptance scoring rubric and eligibility facts.
+- Any other installed ecosystem / validation / branding / investor-prep skills.
+
+Distill them into one **unified intelligence model**: the acceptance patterns,
+product/technology/founder/growth/enterprise/AI/physical-AI patterns, and the
+success vs. rejection factors that all the skills point at. Use that model — plus
+the baked-in knowledge below — to generate.
+
+## Research — baked in (default) · live RFS/funded crawl OPTIONAL
+
+The durable NVIDIA Inception knowledge is **embedded below — no research needed for
+the basics** (saves tokens). Recency matters for *this* skill, so **optionally**
+crawl current signals when fresh proof helps: NVIDIA Inception site & eligibility;
+startup showcases/spotlights; partner ecosystem & AI roadmap; robotics / Physical AI
+/ Omniverse / Cosmos / NIM / NeMo / CUDA / AI Enterprise initiatives; recently
+funded/accepted Inception startups; VC-backed, YC, top accelerator, enterprise-AI
+and deep-tech startups; founder interviews; and AI-infra / Physical-AI / robotics /
+enterprise-software trend reports. When you crawl, cite dated source URLs and mark
+unverifiable claims `[UNVERIFIED]`. Extract acceptance / business / technology /
+founder / growth / product / enterprise / infrastructure / AI / physical-AI patterns
+and the common success vs. rejection factors.
+
+### Baked-in knowledge base (default ground truth)
+
+- **Eligibility:** free, equity-free, rolling; the company must be **incorporated**,
+  **< ~10 years** old, have **≥1 developer**, a **working public website**, and a
+  **business email**, and build an **AI / ML / data-science / accelerated-computing
+  product** (not consulting/outsourcing). **Excluded:** crypto mining/trading, pure
+  cloud reselling, reseller/distributor, thin "AI wrappers" with no real compute.
+- **What gets accepted:** GPU-essential deep-tech core; authentic NVIDIA-platform
+  use that drives compute demand; a physical/agentic/enterprise AI lean; a
+  defensible moat (proprietary data/model/hardware/integration); a real, scalable
+  market; credible technical founders; research-to-product trajectory.
+- **Strategic pull (2024–2026):** Physical AI & robotics · Agentic AI · Digital
+  twins & simulation · Sovereign/enterprise AI · Healthcare/biotech · Manufacturing/
+  industrial · Climate/energy · Cybersecurity · Scientific computing.
+- **Ecosystem map (tech → use):** CUDA/cuDNN, TensorRT (fast inference), Triton/NIM
+  (serving), RAPIDS (GPU data), NeMo (LLMs/RAG), Riva (speech), Metropolis (vision),
+  Holoscan (sensor/edge), Omniverse (digital twins), Cosmos (world models), Isaac
+  (robotics), Jetson (edge), DGX/HGX (training), NVIDIA AI Enterprise (platform);
+  domain stacks: BioNeMo, Clara, DRIVE, Morpheus, Earth-2.
+- **Rejection signals:** GPU only nice-to-have; AI-wrapper; consulting; crypto; pure
+  reselling; vague market; no technical depth/differentiation.
+
+## Startup creation rules
+
+**Every startup MUST be:** AI-first · technology-first · scalable · defensible ·
+global · enterprise-ready · VC-fundable · accelerator-ready · NVIDIA-ready.
+
+**It MUST NOT be:** generic SaaS · a copycat · low-tech · a lifestyle business · a
+small local business · dropshipping · a simple marketplace clone · a weak AI wrapper.
+
+## Preferred categories (pick from these)
+
+Agentic AI · Physical AI · Enterprise AI · AI Infrastructure · Developer Tools ·
+Digital Twins · Robotics · Industrial AI · Healthcare AI · Climate AI ·
+Cybersecurity AI · Computer Vision · Scientific AI · Autonomous Systems · AI
+Operating Systems · Knowledge Intelligence · Synthetic Data · Simulation Platforms ·
+Multimodal AI · AI Workflow Automation.
+
+## NVIDIA alignment engine
+
+**Maximize:** GPU utilization potential · CUDA opportunities · NVIDIA ecosystem
+integration · enterprise adoption · technical depth · data flywheel · defensibility
+· AI innovation · research potential · infrastructure relevance.
+
+**Boost if possible (only where authentic):** NIM · NeMo · Omniverse · Cosmos · Riva
+· TensorRT · Digital Twins · Physical AI · Robotics · Simulation · AI Infrastructure
+· Autonomous Systems.
+
+## Founder-quality engine
+
+Create startups that can: realistically qualify for NVIDIA Inception · attract
+venture capital · support multiple products · support long-term ecosystem growth ·
+be expanded by the Product Inventor · be strengthened by the Idea Booster · scale
+internationally · generate enterprise revenue · leverage accelerated computing ·
+become category leaders.
+
+## Name + domain engine (mandatory)
+
+**Name rules:**
+- **One or two words**, **≤ 8 letters total** (excluding spaces).
+- Easy to pronounce · globally brandable · memorable · premium · founder-friendly ·
+  works for enterprise customers.
+- **No numbers, no hyphens**, no awkward spellings; avoid known trademarks where
+  possible.
+
+**Domain rules (verify before recommending):**
+- **Check `.com` availability** for each candidate before finalizing. Use an
+  available tool — e.g. `whois <name>.com`, a DNS lookup (`dig +short <name>.com` /
+  `nslookup`), or a domain-availability web check.
+- **Reject any name without an available `.com`.** Prioritize **clean, exact-match**
+  `.com` domains.
+- If you genuinely cannot verify, mark the domain `[UNVERIFIED — confirm <name>.com]`
+  and provide 2–3 backup names, but never present an unverified domain as available.
+- Record the check (tool used + result + date) in each idea's **Domain Availability
+  Analysis** section.
+
+## Quality thresholds (the gate — 0–100)
+
+Self-score every startup and ship only those that meet **all** minimums (be a tough
+reviewer; justify each score):
+- NVIDIA alignment ≥ **90**
+- Scalability ≥ **90**
+- Defensibility ≥ **85**
+- AI depth ≥ **90**
+- Investor interest ≥ **85**
+
+If a concept misses any threshold, **redesign or discard it.** Quality over quantity.
+
+## Process
+
+1. Build the unified intelligence model (pre-analysis + baked-in KB; optional crawl).
+2. Pick target categories + a sharp **white-space wedge** within a strategic area.
+3. Generate candidate **names** and **verify `.com`** for each; keep only clean ones.
+4. Design the startup against the creation rules, alignment engine, and
+   founder-quality engine.
+5. Self-score against the 0–100 thresholds; redesign/discard until all pass.
+6. **Write every idea file** (below) and hand off to the downstream skills.
+
+Generate **3–6 startup concepts** by default (or a focused single startup if the
+user asks for one).
+
+## Output — write ALL files to `nvidia-inception/`
 
 Create `nvidia-inception/` in the project root if missing, then write:
 
-1. **`nvidia-inception/rfs-research.md`** — the crawl digest:
+1. **`nvidia-inception/founder-intelligence.md`** — the unified intelligence model:
+   patterns absorbed from the sibling skills + baked-in KB, plus (if crawled) ranked
+   hot themes, recently funded/accepted analogs (with dated sources), and white
+   space. Mark `[UNVERIFIED]` where applicable.
 
-```
-# NVIDIA RFS & Funded-Startup Research
-Generated: <YYYY-MM-DD>
+2. **`nvidia-inception/eligibility.md`** — the eligibility checklist + go/no-go gate
+   applied to every concept.
 
-## Sources crawled (with URLs + access date)
-## Requests for Startups / NVIDIA stated priorities   (ranked hot themes)
-## Recently funded / newly accepted Inception startups (table: name · what · round + date · NVIDIA tech · vertical · source)
-## Common acceptance patterns (traits funded/accepted startups share)
-## White space (under-served angles inside hot themes)
-## [UNVERIFIED] / gaps
-```
+3. **`nvidia-inception/startup-ideas/_index.md`** — the locked target categories,
+   then a table of every shipped startup: name, `.com` status, category, one-liner,
+   primary NVIDIA hardware, and the five threshold scores (all ≥ minimums).
 
-2. **`nvidia-inception/eligibility.md`** — the eligibility checklist + go/no-go
-   gate used for every idea.
+4. **`nvidia-inception/startup-ideas/<StartupName>.md`** — one full file **per
+   startup**, using the **required output** sections below.
 
-3. **`nvidia-inception/startup-ideas/_index.md`** — locked hot themes, then a
-   table of every idea: name, one-liner, RFS theme, funded analog, primary NVIDIA
-   hardware, and **Acceptance Chance 10/10**. Only list dual-clean 10/10 ideas.
-
-4. **`nvidia-inception/startup-ideas/<IdeaName>.md`** — one file **per idea**
-   (brandable name, 4–10 letters, letters only), each:
+### Per-startup file — required output (all sections)
 
 ```
-# <IdeaName>
+# <StartupName>            (one/two words, ≤8 letters, no numbers/hyphens)
 Generated: <YYYY-MM-DD>
-One-liner: <sharp, NVIDIA-aligned, no hype>
-
-## RFS / theme it targets        (link to a Step 1A signal)
-## Recently funded analog         (a Step 1B startup proving the space is live + source)
-## White-space wedge              (how this is differentiated, not a clone)
-
-## Problem & target user
-## Solution (deep-tech product, how it works)
-## Why GPU / accelerated compute is essential
-## Physical AI / Agentic angle    (if applicable)
-## Defensibility / moat (data/model/hardware/integration flywheel)
-## Market: beachhead → expansion → TAM, and why now
-
-## NVIDIA Stack (physical → SDK → software, each tied to a real need)
-### Physical / compute hardware (GPUs first)
-### SDKs / libraries
-### Software / platforms
-
-## Eligibility fit                 (incorporated, <10yr, ≥1 dev, website, business email, AI product, not excluded)
-## Acceptance Chance Scorecard — 10/10
-| Criterion | Score | Why |
-|---|---|---|
-| Deep tech / IP (0–2) | | |
-| GPU-essential (0–2) | | |
-| Physical/Agentic AI (0–1) | | |
-| Defensible moat (0–2) | | |
-| Market & why-now (0–1) | | |
-| NVIDIA ecosystem + RFS fit (0–1) | | |
-| Eligibility clean (0–1) | | |
-| **Total** | **10/10** | |
-
-## Proof plan (prototype → benchmark → pilot → publication)
-## Risks & honest gaps             (incl. [UNKNOWN]/[UNVERIFIED])
-## Next steps                      (hand to nvidia-inception-idea-booster, then -starter, then -product-inventor)
+
+## Startup Name             (and why it fits the name rules)
+## Domain Availability Analysis   (<name>.com check: tool + result + date; backups if needed)
+## Category                 (from preferred categories)
+## Industry
+## Vision Statement
+## Problem Statement
+## Solution
+## Target Market
+## AI Components            (the AI engines; real vs [ASPIRATIONAL])
+## Physical AI Opportunities
+## NVIDIA Ecosystem Opportunities   (specific tech → real need)
+## Technical Architecture
+## Revenue Model
+## Scalability Analysis     (incl. score /100, must be ≥90)
+## Competitive Advantage
+## Defensibility Strategy   (moat; incl. score /100, must be ≥85)
+## Data Flywheel
+## Expansion Opportunities
+## Product Inventor Compatibility   (how it's ready to expand into a 4–6 product ecosystem — do NOT build it here)
+## Idea Booster Compatibility       (how it's ready to be elevated — do NOT boost it here)
+## NVIDIA Inception Eligibility Review   (each criterion: pass + why)
+## Acceptance Probability Analysis      (band + the few strongest acceptance drivers)
+## Investor Appeal Analysis              (incl. score /100, must be ≥85)
+## Long-Term Growth Potential            (path to category leadership)
+
+## Threshold Scorecard (all must pass)
+| Metric | Score /100 | Min | Pass? | Why |
+|---|---|---|---|---|
+| NVIDIA alignment | | 90 | | |
+| Scalability | | 90 | | |
+| Defensibility | | 85 | | |
+| AI depth | | 90 | | |
+| Investor interest | | 85 | | |
+
+## Handoff next steps   (→ Idea Booster → Product Inventor → Starter)
+## Risks & honest gaps  (incl. [UNKNOWN]/[UNVERIFIED])
 ```
 
 ## Quality bar
 
-- **Evidence over opinion.** Every theme and funded analog cites a real, dated
-  source URL. Unverifiable claims are marked `[UNVERIFIED]`, not stated as fact.
-- **Recency matters.** Prioritize signals from the **last 6–12 months**; note the
-  date of each so the research can be refreshed.
-- **Dual gate:** ship only ideas that are **10/10 acceptance chance** AND clear
-  **eligibility** with zero disqualifiers.
-- **Real, buildable ideas** — deep-tech and GPU-essential, never AI wrappers,
-  consulting, crypto, or reseller plays.
-- Names obey **4–10 letters, letters only**; one file per idea; folder/file names
-  match the idea name.
+- **Foundation only.** Create the startup + files; don't do the Booster's,
+  Inventor's, or Starter's work — set them up to succeed.
+- **All thresholds met.** Every shipped startup scores ≥ the five minimums, with
+  justified scores; redesign or discard the rest.
+- **Real `.com`.** Every recommended name has a verified-available (or clearly
+  `[UNVERIFIED]`-flagged) `.com`; names obey the ≤8-letter / no-number / no-hyphen rule.
+- **NVIDIA-ready & GPU-essential.** Authentic ecosystem fit tied to real needs;
+  trips zero disqualifiers (no AI wrappers, consulting, crypto, reseller).
+- **Elevate, don't fabricate.** Mark aspirational items `[ASPIRATIONAL]` and unknowns
+  `[UNKNOWN]`; cite dated sources for any live-crawled claim.
 
 ## Verification checklist (run before finishing)
 
-1. Crawled NVIDIA RFS/priorities AND recently funded/accepted Inception startups;
-   listed sources with URLs + dates in `rfs-research.md`.
-2. Extracted ranked hot themes, common acceptance patterns, and white space.
-3. `eligibility.md` exists with the full checklist and a go/no-go gate.
-4. 3–6 idea files exist, each mapping to a real RFS theme **and** a recently
-   funded analog (with source), attacking distinct white space.
-5. Every idea is **10/10 acceptance chance** (scorecard sums to 10) and clears
-   eligibility with zero disqualifiers; weaker ideas were redesigned or dropped.
-6. Each idea is GPU-essential with an authentic NVIDIA stack (physical → SDK →
-   software) and a believable moat.
-7. `_index.md` lists only the 10/10 ideas with their RFS theme + funded analog.
-8. All names are 4–10 letters (verified); `[UNKNOWN]`/`[UNVERIFIED]` used instead
-   of invented facts.
+1. Pre-analysis done: read the sibling skills and built the unified intelligence model.
+2. Did NOT perform the Booster/Inventor/Starter tasks — only founded the startup(s).
+3. `founder-intelligence.md` and `eligibility.md` exist; live-crawled claims are dated/cited.
+4. **3–6 startups** (or the requested single one), each in its own file under
+   `startup-ideas/`, with **all required sections**.
+5. Every name is **one/two words, ≤8 letters, no numbers/hyphens**, and each has a
+   **`.com` availability check** recorded (available, or `[UNVERIFIED]` with backups).
+6. Every startup meets **all five 0–100 thresholds** (scorecard) and clears
+   eligibility with zero disqualifiers; weaker concepts were redesigned/dropped.
+7. Each startup is GPU-essential with authentic NVIDIA ecosystem fit and a believable moat.
+8. Each file includes Product Inventor + Idea Booster **compatibility** notes (handoff-ready).
+9. `_index.md` lists only shipped startups with `.com` status + the five scores.
+10. No invented facts; `[UNKNOWN]`/`[UNVERIFIED]`/`[ASPIRATIONAL]` used as needed.