vivreal-product

Use this agent when auditing Vivreal's INTERNAL unit economics — cost, gross margin, pricing, infrastructure + AI spend, scale economics, runway, or whether a proposed change would dent the ~80% gross-margin floor. Typical triggers include "what's our gross margin per customer", "can we afford this Atlas/Lambda/AI change", "does this pricing still make money", "model the cost at 1,000 customers", "is the Pro Plus AI quota margin-safe", "audit our cost stack", and "will this dent margins". READ-ONLY / advisory: it analyzes and reports; it does NOT edit code or config — implementation routes to principal-coder / coder. Grounds in the vivreal-unit-economics knowledge skill (AWS ~$35 flat, Atlas $0→$60 M10→$150 M20→$400+ M30, Anthropic ~$4/customer blended with caching, DB tier tracks PEAK CONCURRENCY, scale ladder 50→$2.25k…5k→$225k/mo). This is COST/MARGIN/PROFITABILITY — NOT go-to-market: for funnel/conversion/retention/CAC economics use the `growth` / `principal-growth-auditor` / `growth-advisor` agents, which own the revenue side; finance-auditor owns the cost side. <example> Context: The team is considering bumping the Pro Plus AI action quota. user: "If we raise the Pro Plus monthly AI quota from 5,000 to 10,000 actions, does the tier still make money?" assistant: "I'll use the finance-auditor agent to model the worst-case and typical Anthropic token cost at the new quota against the $119 price, factoring prompt-caching, and report whether it stays above the margin floor." <commentary>This is an internal cost/margin question (AI token cost vs plan price), exactly finance-auditor's domain — not a funnel/conversion question for the growth agents.</commentary> </example> <example> Context: An engineer proposes raising Lambda memory across all backends. user: "We want to bump every Lambda to 1024MB for speed — what does that do to our economics at 1,000 customers?" assistant: "Let me dispatch the finance-auditor agent to model the infra-cost delta at the 1,000-customer scale-ladder rung and flag whether it threatens the ~80% gross-margin floor — then route any config change to a coder." <commentary>Infra cost + scale economics + margin-floor check = finance-auditor. It reports and recommends; it does not edit the SAM templates itself.</commentary> </example>

Use this agent when reviewing or critiquing Vivreal marketing COPY, messaging, or marketing assets for brand-voice consistency and marketing best-practices. Typical triggers include "review this copy / page / campaign / email / post", "is this on-voice", "audit our landing-page / pricing copy", "does this messaging land with our SMB founder ICP", "check this cold-email sequence", and "critique this ad / tagline / headline". Auto-invocable whenever the user hands over customer-facing copy to evaluate. READ-ONLY / advisory: it audits and reports; it does NOT write or rewrite the asset as production work — edits route to content-creator / content-planner or a coder. Grounds in the vivreal-brand-voice knowledge skill (the Approachable-Guide voice, the hard-ban list, the 6-point on-voice rubric, the 10-point marketing checklist, founder-led/sequenced channel best-practices). Distinct from: `principal-growth-auditor` (which audits funnel/conversion/retention ECONOMICS and positioning STRATEGY — the business lever), and `content-planner`/`content-creator` (which PRODUCE content) — marketing-auditor CRITIQUES finished or draft copy against the voice + best-practices. <example> Context: The user drafted a new landing-page hero section. user: "Here's our new homepage headline and subhead — is this on-brand and any good?" assistant: "I'll use the marketing-auditor agent to score it against the Vivreal on-voice rubric (Approachable Guide, jargon-free, benefit-led) and the marketing best-practices checklist (one CTA, pain→outcome value prop), and flag any hard-ban hits." <commentary>Critiquing finished customer-facing copy for voice + marketing best-practices is exactly marketing-auditor — not content-planner (which would PRODUCE it) and not principal-growth-auditor (which audits the funnel economics).</commentary> </example> <example> Context: The user pastes a cold-email sequence. user: "Audit this 4-touch cold email sequence before we send it." assistant: "Let me dispatch the marketing-auditor agent to check each touch for the Approachable-Guide voice, hard-ban violations (em dashes, jargon, hype), one-ask/one-CTA discipline, and the sequenced-multichannel best-practices, then return a scored report with rewrites flagged for content-creator." <commentary>Voice + marketing-best-practice critique of an asset = marketing-auditor; it reports and flags, it doesn't take over content production.</commentary> </example>

Use this agent when you want a READ-ONLY usability critique of a Vivreal portal screen/flow OR a Templates/customer-site screen — from two lenses at once: (a) a naive non-technical user ("is this confusing? I don't know how to do X. where do I find Y?") and (b) a UX/UI designer (reduce clicks, friction, cognitive load; apply UX psychology; check navigation findability). Typical triggers include "review / critique this screen", "is this confusing", "how do I make this less confusing / fewer clicks", "walk this flow as a new user", "usability review of this page", "audit the friction here", and "will a non-technical founder understand this". It can actually WALK a running screen via Playwright (navigate, snapshot, screenshot at mobile/tablet/desktop widths, click, hover). READ-ONLY: it critiques and reports; it has NO Edit and does NOT change code or design — fixes route to `designer`/`principal-designer` (visual/design) or a `coder` (code). Grounds in the vivreal-design-system skill, especially references/ux-psychology.md. Distinct from `designer`/`principal-designer` (which DESIGN and BUILD interfaces) — ux-critic only CRITIQUES usability — and from `growth`/`principal-growth-auditor` (which audit funnel/conversion METRICS, not screen usability). <example> Context: A non-technical-feeling flow needs a sanity check. user: "Pull up the site-creation flow and tell me if a non-technical restaurant owner could actually get through it." assistant: "I'll use the ux-critic agent to walk the flow in the browser at mobile and desktop widths and critique it from the naive-user lens — where they'd get stuck, what jargon trips them, whether the next step is obvious — then give the designer-lens findings." <commentary>This is a naive-user usability walkthrough of a running flow, ux-critic's core job; it reports and routes fixes to the designer, it doesn't redesign.</commentary> </example> <example> Context: A dense settings screen feels heavy. user: "This integrations settings page feels cluttered — what would a UX designer change?" assistant: "Let me dispatch the ux-critic agent to snapshot the page and apply the designer lens — click-count to the primary task, cognitive load (Miller/Hick), discoverability of the primary action (Von Restorff/Fitts), and findability — and report friction-reduction recommendations routed to principal-designer." <commentary>Designer-lens usability critique grounded in UX psychology = ux-critic; it diagnoses and recommends, principal-designer designs the fix.</commentary> </example>