Productised MCP server icon

Productised

by Productised

HIPAA CompliantSOC2 ReadyISO 27001 Ready
Sales & CRM73 tools

Build, brand, publish and analyse AI-powered assessments and lead magnets from an AI agent. 73 tools — 34 read, 35 write, 4 delete — enumerated identically by Anthropic and by Productised. OAuth 2.0 with read, write and offline_access scopes.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

Connect Productised via MCP

https://mcp.productised.ai/mcp

Works in any MCP-compatible client. In Agentman Studio it is one click — no config file to edit.

Productised Tools & Capabilities (73)

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • write is one grant over 39 mutating tools. The scopes are real, but no boundary falls between the routine write and the irreversible one. Granting the connector permission to tag a lead grants it permission to run delete_product.
  • The demo URL is public and cannot be turned off. Every write returns a ?demo=true link that Productised documents as bypassing authentication, domain locks and submission caps. Productised answers "Can I disable demo URLs?" with "Not currently", and notes that revoking your API key stops new ones being generated but leaves existing ones live for as long as the product exists. Security rests on the URL being long and unguessable, which is a different property from access control.
  • delete_product deletes the leads with the product. Productised documents it as permanently removing the product and all of its responses, pages and digital-card links. It requires confirm:true; the other three delete tools have no documented confirmation step.
  • set_webhook has no getter. Productised states the current webhook is visible only in the app's product settings, so an agent cannot read back a destination it — or a previous session — configured.
  • Document mode executes JavaScript on public pages. Saved HTML beginning <!DOCTYPE html> renders in an iframe where, per Productised, scripts execute and page JS can read respondent data via window.PRODUCTISED_DATA. Script-stripping applies only to fragment mode.
  • We could not read tool schemas or safety annotations. The endpoint returned 401 to an anonymous request on 2026-08-23, so no parameter-level detail or readOnlyHint is published here.
  • The connected-MCP surface is unbounded and runs unsupervised. add_mcp_connection attaches an arbitrary external server whose tools the product's AI can call mid-conversation with a member of the public. Nothing in the 73-tool count reflects it.
  • No SSRF guard is documented for add_brain_source. Productised documents thorough guards for upload_image but describes add_brain_source only as fetching and storing a URL's content. We found no statement either way; treat it as unverified rather than unguarded.
  • No per-call credit rate is published. Productised states consumption varies with the complexity of the operation, so the cost of a generate_product or brain_ask cannot be predicted before the call. It also publishes no idempotency guidance, so a generate_product that times out should not be blindly retried.
  • revise_product is documented as lossy, replacing the entire workflow rather than editing it. Productised recommends update_node for surgical changes.
  • Single-user only. Productised states there are no team seats or multi-user invitations on any plan, so there is no role separation to restrict what a connected agent may reach.
  • Some capabilities are plan-gated. Custom branding, custom domains and page SEO start at Starter; AI Insights, Talking Points and BYOK are Pro-only. get_product_insights and get_talking_points therefore do not function on lower tiers.
  • The widget surface is unaudited. Anthropic marks the entry has_mcp_app: true. We did not test the widget, and its behaviour is not derivable from the tool list or the scopes.

Frequently asked questions

It builds and publishes lead-capture assessments. Productised is a hosting platform for AI-powered quizzes, scorecards and diagnostics that qualify visitors and collect their contact details. The connector lets an agent create a product, write its questions and scoring, design its result page, point a custom domain at it, publish it to a public URL, and then read the leads it captures.

Seventy-three, of which 39 change something. Our classification of the names against Productised's own tool reference is 34 read, 35 write and 4 delete. The four deletes are delete_product, delete_page, delete_node and delete_brain_item. Only delete_product is documented as requiring a confirm:true flag, and it removes the product with all of its responses.

Yes, and that is its purpose. publish_product sets a product to public and returns a live shareable URL. set_custom_domain and set_workspace_domain point your own hostname at it, and set_welcome_html and update_page write the HTML people land on. Treat every write here as potentially reaching an audience rather than an internal record.

Three: read, write and offline_access. Productised's authorization server advertised those on 2026-08-23. They are real application scopes rather than identity claims, but write is a single grant covering all 35 write tools and all 4 delete tools, so no consent choice separates tagging a lead from deleting a product.

No. Productised documents that responses are immutable through the connector, with no tool to delete or edit a lead's answers, to protect the audit trail. tag_response and update_response_status change only pipeline metadata. The exception is indirect: delete_product removes a product and all of its responses together.

It spends AI credits, not money. Productised documents that generate_product, revise_product, generate_plan and brain_ask run AI pipelines that consume your plan's monthly credit allocation. Three tools observe the spend — get_credits, get_plan_status and get_workspace_stats — and a blocked write returns a structured limit_reached payload rather than billing you extra.

Anyone with the link. Productised documents that every write tool returns a ?demo=true preview URL that is publicly accessible and bypasses authentication, custom-domain locks and submission caps. Productised describes it as long and non-enumerable rather than access-controlled, says it cannot currently be disabled, and confirms existing demo URLs stay live even after you revoke the API key.

Almost always a popup blocker. Productised's troubleshooting page says to allow popups for claude.ai and click Connect again. If the popup opens but sign-in fails, you may be signed in to a different Productised account, or your organisation may block third-party OAuth grants — in which case Productised documents an API-key route through Claude Desktop's JSON config as the fallback.

No. No tool sends mail, and Productised states the connector cannot read or send email or messages. The outbound tools are named for a campaign but only build strings: generate_outbound_urls_from_csv takes your prospect CSV and returns the same CSV with an outbound_url column added. You send it from your own mail-merge tool.

Sources

Use in Agentman

Connect once and your agents call these tools on their own — on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio

Resources

Server Info

Category
Sales & CRM
Developer
Productised
Tools
73
Domain
mcp.productised.ai

Using Claude Desktop or another MCP client? Setup docs — the connection URL above works anywhere.

Similar MCP Servers for Sales & CRM

HubSpot logo

HubSpot

by HubSpot

Coming Soon
Sales & CRM5 tools
View detailsConnect
ZoomInfo logo

ZoomInfo

by ZoomInfo

Coming Soon
Sales & CRM10 tools
View detailsConnect
Explorium logo

Explorium

by Explorium.ai

Sales & CRM10 tools
View detailsConnect

Related Resources

Browse All MCP ServersClaude & ChatGPT Setup GuideBrowse AI Skills