# Inboxmate Batch Demo

> Batch-create InboxMate demos for CRM prospects. Queries Twenty CRM for companies without opportunities, validates their websites, creates demos for valid ones, and marks unreachable/outdated ones as DISQUALIFIED. Optional parameter: track ('inbox' builds Demo-Postfächer via /inboxmate-inbox-demo, 'chatbot' builds chatbot demos; default reads the Track from each company's qualification note).

- Skill: `psquared-development/inboxmate-batch-demo` (Agent Skill)
- Install (CLI): `npx skillmds@latest add psquared-development/inboxmate-batch-demo`
- Raw SKILL.md: https://api.skillmd.com/api/skills/psquared-development/inboxmate-batch-demo/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: psquared-development (https://skillmd.com/u/psquared-development)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/psquared-development/inboxmate-batch-demo

---


# InboxMate Batch Demo Pipeline

## Track routing (which demo to build)

`/inboxmate-batch-demo [track]` — `track` is optional:

- **`inbox`** — build a Demo-Postfach for every company in the batch via `/inboxmate-inbox-demo` (no agent, no widget; opportunity gets `demoType: INBOX`).
- **`chatbot`** — build chatbot demos via `/inboxmate-demo` for every company (`demoType: CHATBOT`).
- omitted — read each company's qualification note (`Track: CHATBOT | INBOX` set by `/find-leads`) and route per company; if no Track noted, default to CHATBOT.

## Prerequisites — Environment Variables

All required tokens are in the **`.env` file in the current working directory** (the agenthub repo root). Read it at startup to get:

- `PSQUARED_CRM_TOKEN` — Bearer token for Twenty CRM GraphQL API
- `NUXT_MCP_DEMO_TOKEN` — Bearer token for the InboxMate MCP server
- `OPENBRAND_API_KEY` — API key for OpenBrand brand color/logo extraction

**Do this first:** Read `.env` from the current directory and extract these values. If any are missing, stop and ask the user.

**Before starting, ask the user for the offer deadline:**

> ```
> When should the demo offers expire?
> This sets the countdown timer on all demo pages in this batch.
> Examples: "in 14 days", "2026-04-01", "end of month"
> ```

Wait for the answer. Convert to an ISO 8601 date. Pass this deadline to every demo created in the batch — do NOT ask again per company.

> **Announce:**
> ```
> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
> InboxMate Batch Demo Pipeline
> Offer deadline: [date]
> Reading .env for CRM and MCP tokens...
> Querying CRM for unprocessed prospects...
> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
> ```

---

## Step 1 — Get Prospects from CRM

Read `PSQUARED_CRM_TOKEN` from `.env` and use it to query the Twenty CRM (GraphQL at `https://crm.psquared.dev/graphql`) for companies that do NOT yet have an opportunity. Use this approach:

1. Fetch companies where `outreachFor` contains `INBOX_MATE` (up to 100)
2. Fetch all opportunities (up to 200)
3. Filter out companies that already have ANY opportunity (regardless of stage)

The remaining companies are unprocessed InboxMate prospects.

**Critical filter:** Only fetch companies tagged for InboxMate outreach. Never process `PERSONAL_ONLY` companies — these are Martin's private contacts. Companies tagged `PSQUARED_SERVICES` only are handled by `/find-services-leads` and its own funnel, not this pipeline. Companies tagged `INBOX_MATE_MUSEUM` only (the curated museum list) are also excluded automatically — they get a separate museum-specific outreach flow when one is built.

```bash
curl -s -X POST https://crm.psquared.dev/graphql \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $PSQUARED_CRM_TOKEN" \
  -d '{"query":"{ companies(filter: { outreachFor: { containsAny: [INBOX_MATE] } }, first: 100) { edges { node { id name domainName { primaryLinkUrl } outreachFor } } totalCount } }"}'
```

> **Legacy companies (pre-field):** Companies created before the `outreachFor` field existed have `outreachFor: null` and will NOT match this filter. That is intentional — they need to be tagged before automation touches them. If the user wants to process legacy untagged companies, they should tag them first (or we do it under explicit instruction).

> **Announce:** `Found [N] unprocessed prospects.`

If there's a specific company or list the user wants to process, use that instead.

---

## Step 2 — Validate Each Prospect's Website

For each prospect, **before running the demo pipeline**:

### 2a — Check if the website is reachable

Use `WebFetch` on the company's domain (from CRM `domainName.primaryLinkUrl`).

**Skip the prospect if ANY of these are true:**
- Domain returns HTTP error (4xx, 5xx) or times out
- Domain redirects to a parked/expired domain page
- Domain shows a "coming soon" or "under construction" placeholder
- Website has no meaningful content (just a logo and "contact us")
- Website is clearly outdated — copyright year 2+ years old, broken images, obviously abandoned
- Website is a social media profile only (LinkedIn, Facebook) — not a real business site

### 2b — Mark skipped prospects in CRM

**Do NOT ask the user what to do.** Auto-skip and mark in CRM immediately.

First, check if the Company object has a field like `idealCustomerProfile` (ICP) or a rating/status field you can use. Run this introspection query once at the start:

```graphql
query { __type(name: "Company") { fields { name type { name kind ofType { name } } } } }
```

**If there's a usable boolean/enum field** (e.g. `idealCustomerProfile`): set it to `false` or the "not a fit" value to mark the company directly.

**Regardless**, also create a DISQUALIFIED opportunity so the company is excluded from future batch runs:

```graphql
mutation CreateOpportunity($data: OpportunityCreateInput!) {
  createOpportunity(data: $data) { id name stage }
}
```

Variables:
```json
{
  "data": {
    "name": "[Company] — Website not suitable",
    "stage": "DISQUALIFIED",
    "companyId": "[companyId]"
  }
}
```

> **Announce for each skip:** `SKIP: [Company] — [reason]`
> **Never pause to ask.** Just mark and move to the next prospect.

### 2c — Collect valid prospects

Prospects with a working, current website proceed to Step 3.

> **Announce:** `[N] prospects ready for demo creation, [M] skipped.`

---

## Step 3 — Create Demos for Valid Prospects

For each valid prospect, invoke the `/inboxmate-demo` skill (or follow the inboxmate-demo SKILL.md pipeline).

Process **one prospect at a time** — do not parallelize MCP calls.

After each demo is created:
1. The inboxmate-demo pipeline creates an opportunity in CRM at `SCREENING` / `PENDING_REVIEW` with the demo URL — verify it was created
2. Announce the result

> **Announce after each:**
> ```
> DONE: [Company] — [playgroundUrl]
> ```

---

## Step 4 — Summary

After all prospects are processed:

```
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
BATCH COMPLETE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Demos created: [N]
  - [Company 1] → [url]
  - [Company 2] → [url]

Skipped (website issues): [M]
  - [Company A] — [reason]
  - [Company B] — [reason]

Already processed (had opportunity): [K]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
```

---

## Skip Criteria Quick Reference

| Signal | Action |
|--------|--------|
| HTTP error / timeout | SKIP — "unreachable" |
| Parked / expired domain | SKIP — "parked domain" |
| "Coming soon" / "Under construction" | SKIP — "placeholder page" |
| No real content (just logo + contact form) | SKIP — "no meaningful content" |
| Copyright year 2+ years behind current | SKIP — "outdated (copyright YYYY)" |
| Broken images, dead links, 90s design | SKIP — "outdated/abandoned" |
| Social media profile only | SKIP — "no website (social only)" |
| No domain in CRM | SKIP — "no domain on file" |
| Working site with real content | PROCEED with demo |