/diagnostics-intake

name: diagnostics-intake
description: Fast founder-input capture for the Diffmode growth-tactics pipeline (the diagnostics stage). Produces WS/01-diagnostics/founder-input.md in the exact schema the enrichment + synthesis stages read. Two modes — (A) URL mode researches a product's website (homepage/pricing/about) plus a web-research pass to prefill the researchable fields, then marks founder-only gaps for confirmation; (B) Q&A mode formats a minimal (~2-minute) answer set into the schema. Use as the entry point of start, or whenever a workspace needs a founder-input.md before enrichment.
metadata:
  version: "1.0.0"

Diagnostics — Intake (fast founder-input)

You produce a single file: `WS/01-diagnostics/founder-input.md`, in the **exact schema** the rest of the pipeline reads (enrichment, the constraints generator's field parser, and synthesis all key off these sections and labels). This replaces the slow 20-minute diagnostic interview with a ~2-minute path: research what's public, ask only what isn't.

This is **capture, not analysis** — no strategy, no recommendations, no channel picks. That separation is load-bearing: later stages depend on raw, un-editorialized founder context.

Inputs & Output

The invoker provides (do not hardcode absolute paths):

• **MODE A — `url`** (a website, e.g. `https://theona.ai`): research the site + company.
• **MODE B — `answers`**: a block of founder answers to the minimal question set below

(the orchestrator collects these in the main thread; you format them).

• **MODE A+B**: both — research the URL AND fold in any founder answers the brief passed

(answers always win over researched guesses).

• **OUTPUT**: write to `WS/01-diagnostics/founder-input.md` (path supplied by invoker).

If neither `url` nor `answers` is present, write the schema with every must-ask field as a `[NEEDS FOUNDER INPUT: …]` placeholder and report it — do not invent a business.

Field provenance (what to research vs what to ask)

| Field | Provenance | |-------|-----------| | Product description, what it does, who it's for | RESEARCHABLE (homepage/about) | | Business model + pricing (tiers, free trial) | RESEARCHABLE (pricing page) | | Target-audience hypothesis (segments, ICP) | RESEARCHABLE (site copy) + confirm | | Competitive alternatives (direct + indirect) | RESEARCHABLE (web research) | | Product complexity ("explains itself" vs "needs a demo") | RESEARCHABLE + confirm | | **Stage + current metrics** (visitors, signups, MRR, paying customers) | **MUST ASK** | | **Current acquisition sources / what's working** (Q8 demand-gen signal) | **MUST ASK** | | **Demand-gen vs CRO split** (traffic problem vs conversion problem) | **MUST ASK** | | **Budget** (monthly marketing $, paid-ads yes/no) | **MUST ASK** | | **Unfair advantage / rare assets** (technical skill, industry access, network, domain expertise, existing audience) | **MUST ASK** | | **Hours per week for growth** (time they can commit) | **MUST ASK** | | **Goal + timeline** (target metric, deadline) | **MUST ASK** | | Tactics ruled out + competitor-dignity constraints | MUST ASK (optional) |

Founder-only fields cannot be guessed from a website. In URL mode, write a researched best-guess for confirmable fields and a `[CONFIRM: …]` / `[NEEDS FOUNDER INPUT: …]` marker for every MUST-ASK field not supplied in `answers`, and list them all under a final **"Confirmation Gaps"** block so the orchestrator can ask.

Minimal question set (Q&A mode — target ≤ 2 minutes)

These are the ONLY fields a founder must answer (the orchestrator asks them; skip interview-observation and CRO-deep-dive fields):

1. **Product in one line** — what it does + who it's for (skip if URL covers it). 2. **Business model & pricing** — e.g. "B2B SaaS, $49/mo, free trial" (skip if URL covers). 3. **Target-audience hypothesis** — the 1-3 roles/segments you think you're for. 4. **Stage + rough metrics** — pre-launch / early / traction / growth, plus any of: visitors/mo, signups/mo, paying customers, MRR. 5. **Q8 — current acquisition sources** — where today's users/traffic come from, and what's working best. *(This is the demand-gen-vs-CRO signal. If traffic is the bottleneck → demand gen; if traffic converts poorly → CRO.)* 6. **Biggest growth problem** — "not enough traffic" (demand gen) **or** "traffic doesn't convert" (CRO), in the founder's words. 7. **Budget + unfair advantage + hours** — monthly marketing $ (and paid-ads yes/no); what they have that's rare (technical skill / industry access or network / existing audience / domain expertise / none yet); hours per week they can put into growth. 8. **Goal + deadline** — target (e.g. "20 paying teams" / "10k visitors/mo") + timeframe. *(Optional 9th: tactics/channels already ruled out + why.)*

Keep each question answerable in a phrase. Never block on the optional one.

Procedure

Mode A — URL research → prefill

1. **Fetch the site.** Use your page-fetch tool (a web-fetch capability, or a scrape skill if available) on the homepage, the pricing page, and the about/product page. Extract: what the product does, who it's for, business model, pricing tiers, free trial, positioning language. 2. **Research the company + market** with your web-research backend (whatever search + page-retrieval tools your runtime gave you): confirm the category, find direct + indirect competitors/alternatives, note stage signals if public (funding, team size, launch date). Cite URLs + access dates in a Research Notes footer. 3. **Fill researchable fields** from steps 1-2. For confirmable judgment calls (audience hypothesis, product complexity), write your best read AND a `[CONFIRM: …]` marker. 4. **Fold in `answers`** if the brief passed any (founder answers override guesses). 5. **Mark must-ask gaps.** Every MUST-ASK field not covered by `answers` gets a `[NEEDS FOUNDER INPUT: …]` marker inline and an entry in the Confirmation Gaps block.

Mode B