--- name: sales-cresh description: "Cresh (cresh.me) platform help — a multi-agent AI startup/business-idea validator that scores an idea across 33 metrics in 6 groups (market, business model, problem fit, product/tech, go-to-market, feasibility & risks), each metric scored with an explanation, improvement tips, and per-metric web search, then exports a PDF in ~3–6 minutes. Free tier shows ~5–6 of 33 metrics; the full report and PDF are paid via credit-based Cresh Coins (best-effort, confirm on site), plus a separate enterprise 'decision intelligence' tier for innovation teams. Use when running an idea through Cresh, interpreting its 33-metric report or group scores, judging whether its cited sources are trustworthy, understanding Cresh Coins pricing or the free-tier metric limit, exporting the PDF, or asking whether it has an API. UI-only — no public API, webhooks, Zapier, or MCP. Do NOT use for the validate-before-building method or comparing validators (use /sales-idea-validation), or building a smoke-test landing page (use /sales-funnel)." argument-hint: "[describe what you need help with in Cresh]" license: MIT version: 1.0.0 tags: [sales, validation, pre-launch, platform] --- # Cresh Platform Help Cresh (**cresh.me** — the live product; the backlog's `cresh.ai` does not resolve) is a **multi-agent AI startup/business-idea validator**. You describe an idea in a guided **wizard** (problem, solution, audience, competition, business model) and it returns a **structured report scoring the idea across 33 metrics** organized into **6 groups** — Market & Competition, Business Model & Revenue, Customer Need & Problem Fit, Product & Technology, Go-to-Market, and Feasibility & Risks. Each metric gets a **score, a plain-language explanation, and improvement suggestions**, and Cresh markets a **per-metric web search** that pulls "real sources and real data" behind each score. A full run takes **~3–6 minutes** and exports a **PDF** to share with a co-founder or investor. It's a fast, structured second opinion — **not a substitute for real demand validation** — the report is an AI opinion, the "sources" are **disputed** (reviewers report no visible citation trail), and it has **no public API** (UI tool). ## Step 1 — Gather context If `references/learnings.md` exists, read it first for accumulated platform knowledge. Ask only what you can't infer: 1. **What do you want from Cresh?** - A) Run an idea through it and interpret the report — the 33 metrics, the 6 group scores, or a section - B) Understand the credit-based **Cresh Coins** pricing and the **free-tier metric limit** (~5–6 of 33) - C) Judge whether its cited "real sources" can be trusted, or export/share the **PDF** - D) Automate or export programmatically (API reality) 2. **What's the real question — "is this idea worth building?" or "how do I use this tool?"** If it's the go/no-go decision, that's `/sales-idea-validation` (a score is not demand) — route in Step 2. 3. **Are you the target user?** Cresh's self-serve tier is for **solo founders / indie hackers**; its separate **enterprise "decision intelligence"** tier (InnoScore, €500–€10k/mo) is for **innovation teams** — don't send a solo founder to the enterprise offer. Skip-ahead: if the user wants the validate-before-building *method* or to compare validators across the market, route to `/sales-idea-validation` immediately. ## Step 2 — Route or answer directly | If the user's question is about… | Route to | |---|---| | The validate-before-building **method**, or the go/no-go decision itself | `/sales-idea-validation {question}` | | Comparing Cresh against other validators (Preuve, DimeADozen, IdeaScore, Validator AI…) | `/sales-idea-validation {question}` | | Building the smoke-test / fake-door **landing page** to test real demand | `/sales-funnel {question}` | | Growing a pre-launch **waitlist** / capturing real demand | `/sales-audience-growth {question}` | | Turning the report's market/audience sections into a real **content** program | `/sales-content {question}` | When routing, give the exact command: "This is a {domain} question — run: `/sales-idea-validation {original question}`" Otherwise, answer Cresh-specific questions using Step 3. ## Step 3 — Cresh platform reference **Read `references/platform-guide.md`** for the full reference — the wizard input, the **6 metric groups and 33 metrics** (with the labels marketing uses), the per-metric explanation/improvement structure, the **"deep search per metric" claim and why its source transparency is disputed**, the **6-metric free snapshot vs the full 33-metric paid report + PDF**, the **credit-based Cresh Coins** model and the best-effort price points, the **enterprise "decision intelligence" tier** (and why it's off-target for a solo founder), and the **no-public-API** automation reality. Answer using only the relevant section — don't dump the full reference. ## Step 4 — Actionable guidance - **Treat the 33-metric score as a directional gut check, not demand.** The whole report is an **LLM opinion** generated from a short idea description — the group scores, the market sizing, and the competitor set are **AI estimates**, not measured demand. Tell the user to **keep the structured outputs** (the flagged weak metrics, competitor read, risks, improvement tips) to sharpen the pitch, and take the **go/no-go from real behavior** — a smoke test or pre-sale — not from a high Cresh score. Route the real test to `/sales-idea-validation`. - **Don't over-trust the "real sources / real data" claim — verify the citations yourself.** Cresh's differentiator is a per-metric web search that supposedly cites real sources, but **reviewers report no visible citation trail** and market/retention numbers you can't trace. Before quoting any figure or "sourced" claim, **open the underlying source** (if one is shown) and sanity-check it against a primary reference — treat unsourced numbers as AI estimates. This is the report-auditing discipline in `/sales-idea-validation`, applied to Cresh. - **Do the credit math on Cresh Coins, and flag the pricing as unsettled.** Cresh runs on **credit-based "Cresh Coins"**: a **free tier shows only ~5–6 of 33 metrics** (a snapshot), and the **full 33-metric report + PDF** costs coins. Best-effort at research (**sources conflict**): ~$3 per analysis in some reviews, or roughly **$9 / $19 / $29 monthly for 1 / 3 / 5 analyses** in others. **Always tell the user to confirm current pricing on cresh.me** and to finalize the idea framing before spending a coin — garbage-in, garbage-out, and each pivot re-run costs again. - **Match the tier to the user — don't send a solo founder to the enterprise offer.** The self-serve product is for indie founders; the **enterprise "decision intelligence" / InnoScore** tier (€500/mo pilot → €2k–€10k/mo + a large implementation fee, for 200–5,000-employee innovation teams) is a different product. A solopreneur validating one idea wants the self-serve tier. - **There's no public API — don't plan an integration around it.** If asked to automate, batch, or pipe reports elsewhere, say plainly there's **no documented public API, no webhooks, no Zapier/Make, no MCP**. The **only way to get a report out is the PDF export** (or copy/screenshot from the browser) — there is no supported programmatic or batch path. The thing worth automating is the real signal (landing-page conversions), not the generated score — route that to `/sales-idea-validation` and `/sales-funnel`. If you discover a gotcha or tip not in `references/learnings.md`, append it there with today's date. ## Gotchas *Best-effort from research (2026-07) — Cresh's pricing, credit terms, metric groups, and source claims move; verify at cresh.me.* - **Right domain: cresh.me.** The live validator is at **cresh.me**; the backlog's `cresh.ai` **does not resolve** — don't route users there. - **The 33-metric score is an AI opinion, not demand.** A high aggregate score generated from a short description validates nothing on its own — follow it with a real-behavior test (smoke test / pre-sale). - **"Real sources" is disputed.** Cresh markets per-metric web search with real data, but **reviewers report no visible citation trail**; verify any sourced figure against a primary reference before quoting it. - **Garbage-in, garbage-out.** Output quality tracks input specificity — a vague idea yields a generic report, and each pivoted re-run spends another credit. Sharpen the customer + problem + differentiation first. - **It can overlook adjacent segments.** Reviewers note Cresh anchors on the segment you describe and can miss adjacent buyers — don't treat its market read as exhaustive. - **Pricing is credit-based and unsettled.** Cresh Coins fund each run; the free tier shows only ~5–6 of 33 metrics, and quoted prices (per-analysis vs monthly) **conflict across sources** — confirm on cresh.me. - **Enterprise tier is a different product.** The €500–€10k/mo "decision intelligence" / InnoScore tier targets corporate innovation teams, not solo founders — don't conflate it with the self-serve validator. - **No public API, no webhooks, no Zapier/Make/MCP.** It's a UI tool; the report leaves only as a **PDF** — there's no supported way to script, batch, or export programmatically. ## Related skills - `/sales-idea-validation` — The tool-agnostic validate-before-building method + the full validator landscape (use this to actually decide build-or-not; a score is not demand) - `/sales-preuve` — Preuve AI platform help (the source-linked validator that **cites every claim** across 50+ live sources — the credible counterpart to Cresh's disputed "real sources" claim) - `/sales-ideascore` — IdeaScore platform help (a lean, letter-graded triage validator; contrast its terse "decision-ready" output against Cresh's fuller 33-metric report) - `/sales-dimeadozen` — DimeADozen platform help (a one-time 200+ page report validator; contrast its depth + disputed citations against Cresh's 33-metric report) - `/sales-funnel` — Build the smoke-test / fake-door landing page the real demand test runs on - `/sales-audience-growth` — Grow a pre-launch waitlist and capture real demand - `/sales-do` — Not sure which skill to use? The router matches any sales objective to the right skill. Install: `npx skills add sales-skills/sales --skill sales-do -a claude-code` ## Examples ### Example 1: "Cresh scored my idea high across the 33 metrics — should I build it?" **User says**: "Cresh gave my SaaS idea strong scores across most of its 33 metrics and a good overall read. Is that validated — should I start building?" **Skill does**: Explains the 33-metric report is an **LLM opinion** generated from a short description — the scores, market sizing, and competitor set are **AI estimates**, not demand. Tells the user to keep the structured parts (weak-metric flags, competitor read, risks, improvement tips) to sharpen the pitch, then route the real go/no-go to `/sales-idea-validation` — a smoke test or pre-sale where a stranger takes an action — rather than trusting the score. **Result**: The founder earns a real demand signal before building, using Cresh's structure, not its score. ### Example 2: How Cresh Coins pricing and the free tier work **User says**: "How much does Cresh cost? The free run only showed me a handful of metrics — how do I get the full 33 and the PDF, and is it a subscription?" **Skill does**: Explains Cresh is **credit-based Cresh Coins** — the **free tier shows only ~5–6 of 33 metrics** (a snapshot), and the **full 33-metric report + PDF** costs coins. Gives the best-effort price points (~$3/analysis in some reviews vs ~$9/$19/$29 monthly for 1/3/5 in others), **flags them as conflicting/unsettled to confirm on cresh.me**, and advises finalizing the idea framing before spending a coin since each pivot re-run costs again. **Result**: The user understands the metric gate and credit model and confirms live prices before paying. ### Example 3: Can I export or automate Cresh reports via API? (developer/automation) **User says**: "I want to run ideas through Cresh automatically and pull the reports into my own dashboard via an API or Zapier. How do I set that up?" **Skill does**: States plainly there's **no documented public API, no webhooks, no Zapier/Make, no MCP** — it's a UI tool whose only data-out is the **PDF export** (or copy/screenshot), so there's no supported way to batch or export programmatically. Suggests a tool built for programmatic validation if a pipeline is required, and notes the thing worth automating is a real demand signal (landing-page conversions), not a generated score — pointing to `/sales-idea-validation` and `/sales-funnel`. **Result**: The user avoids building on a non-existent API and picks a workable path. ## Troubleshooting ### The scores feel generous / it never really tells me to stop **Symptom**: The 33-metric report reads encouraging even for a weak idea. **Cause**: The report is a short-prompt LLM output — it pattern-matches to plausible-sounding encouragement and can invent supporting "facts" (market size, demand, sources). **Solution**: Stop treating the score as the decision. Keep the structured outputs (weak-metric flags, competitor read, risks, improvement tips) and run a real demand test — smoke test or pre-sale — via `/sales-idea-validation`. Treat a low group score as a cheap early filter, not the go/no-go. ### I can't find the "real sources" behind the metrics **Symptom**: Cresh markets per-metric web search with real data, but you can't see or open the citations. **Cause**: Reviewers report **no visible citation trail** — the "sources" claim isn't consistently backed by traceable references in the report. **Solution**: Treat any figure as an AI estimate until you verify it. Where a source is shown, open it and check it against a primary reference; if you need genuinely cited validation, use a source-linked validator (`/sales-preuve` cites every claim). Never quote a Cresh number you can't trace. ### I want an API or a way to export/automate **Symptom**: Want to script Cresh or pull reports into another system. **Cause**: Cresh has no documented public API, webhooks, or iPaaS connectors; it's a web tool whose only data-out is the PDF. **Solution**: There's no supported programmatic path — export the **PDF** (or copy from the browser) and parse it if you must. Use a tool intended for programmatic use if you need a pipeline, and automate a real demand signal (landing-page analytics) instead. See `/sales-idea-validation` and `/sales-funnel`.