MCP App Store
by KeenEthics
AgentFabrica icon

AgentFabrica MCP Server for Claude and ChatGPT

By AgentFabrica

Overview

AgentFabrica is a marketing operations panel for brands and agencies: content calendar, ad pre-tests, lead tracking and WhatsApp customer conversations in one place. This connector lets Claude read that panel and prepare work for your team. What Claude can do: - List the brands on your account and their tone of voice. - Show the content calendar, upcoming and past posts with their status. - Show everything waiting for a human decision, each with a link to review it in the panel. - Summarize ad pre-tests (panel score, click intent, objections) and how well the panel has predicted this brand's real results. - Report lead counts by status and source, and WhatsApp line status with conversation counts. - Show plan quota usage and recently generated content. - Add a draft post to a brand's approval queue. What it will not do: it never publishes, approves or deletes anything. A draft only goes out after a person approves it in the panel. Lead and WhatsApp tools return counts only, never names, phone numbers or message text. Requirements: an AgentFabrica account on a plan with API access. You sign in with your AgentFabrica account and can disconnect Claude at any time from the panel.

How to connect

  1. Claude: Open the AgentFabrica listing in the Claude connectors directory and choose Connect. Claude listing AgentFabrica exposes 11 tools here.
  2. ChatGPT: Open the AgentFabrica listing in the ChatGPT app directory and choose Connect. ChatGPT listing AgentFabrica exposes 12 tools here.

Tools

ad_performance

ChatGPT
Paid ad performance per channel (meta, google, tiktok, chatgpt, linkedin) for one brand: spend, impressions, clicks, conversions, cost per conversion, revenue and ROAS, each compared with the immediately preceding period of the same length (change covers spend, clicks, conversions, costPerConversion, revenue and roas; it is relative: 0.12 = +12%, -1 = fell to zero, null when the previous value is missing or zero). conversionType is one of purchase, lead, registration, messaging_conversation, conversion, mixed, other. Also returns a total over the measured channels only, a health score, and active alerts (spend spike, spend stopped, cost-per-click spike) computed from the daily series. A channel that is not connected or failed to measure has status not_connected or error and null numbers; it is left out of the total and listed in notMeasured. Amounts are in the ad account currency; when channels use different currencies the total amounts are null (currencyMixed). Counts and totals only: no campaign names, ad names, creatives or account ids. Read-only.

ad_pretests

ChatGPT
Ad pre-tests run before launch: panel score, click-intent rate and the objections the panel wrote. If the ad later ran, the real result is included. The score is the panel's reaction, NOT a reach forecast.

calibration_report

ChatGPT
How well the panel holds up for this brand: in how many comparisons it ranked the ads correctly. With too little data it reports the situation instead of producing a number.

generated_assets

ChatGPT
Generated content, newest first. Optionally filtered to a single brand.

lead_summary

ChatGPT
Lead counts per brand. total, byStatus (new, contacted, appointment, won, lost) and bySource (meta_lead_form, comment_dm, whatsapp, website_form, manual) count every lead recorded for the brand and are not limited to the period. Only newInPeriod (leads created in the period) and fromAdsInPeriod (those of them with ad attribution) use the period. Counts only: no names, phone numbers or messages.

learning_pool

ChatGPT
Quality evidence drawn from the brand's own ad data: which hook, angle, visual style and format works, each with its evidence tier and reasoning. alwaysSeenWith lists the other returned attributes that appeared in exactly the same ads: this data cannot tell which of them made the difference, so treat them as one finding (null when unknown for an older pool). Returns null when no pool exists.

list_brands

ChatGPT
Returns the brands on the account: name, industry, website, tone of voice and brand role (own brand / client brand). Other tools take these ids as brandId.

pending_approvals

ChatGPT
Everything waiting for a human decision: calendar posts in needs_approval plus queued approvals (ads, blog posts, product imports). Each item has a reviewUrl where a person approves or rejects it. This tool cannot approve anything.

quota_status

ChatGPT
The plan's quota status: used, limit, remaining and unlimited for each generation type. Types that cannot be measured are left out.

scheduled_posts

ChatGPT
Posts on the content calendar: title, platforms, time, status (planned, needs_approval, publishing, published, failed, rejected; unknown for an older status the publisher does not recognise, which is never published) and a caption preview. 'upcoming' is soonest first, 'past' is newest first.

submit_post_for_approval

ChatGPT
Adds a draft post to the brand's approval queue. It is NOT published: a person must approve it in the panel (reviewUrl) before it goes out. Platforms: facebook, instagram, linkedin; instagram needs at least one image URL. Without scheduledAt the brand's measured best posting time is used. Submitting the same caption again within 10 minutes returns the existing draft.

whatsapp_summary

ChatGPT
WhatsApp line status per brand (connected, disconnected, pairing, connecting, paused, not_set_up, or unknown when the line reports a status outside this list; QR or official line) and conversation counts: active, unread, needing a human, handled by the assistant, hot leads, new in the last 7 days. Counts only: no numbers, names or message text.

ad_pretests

Claude

calibration_report

Claude

generated_assets

Claude

lead_summary

Claude

learning_pool

Claude

list_brands

Claude

pending_approvals

Claude

quota_status

Claude

scheduled_posts

Claude

submit_post_for_approval

Claude

whatsapp_summary

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/agentfabrica"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/agentfabrica)