Skip to content

PageCRM MCP Server

Connect AI assistants to PageCRM with governed MCP access

PageCRM MCP gives approved AI clients controlled access to CRM tools, resources, and prompt templates. It is designed for sales copilots, support assistants, internal automation agents, and AI systems that need CRM context without bypassing organization boundaries.

Open integrationsRead mcp.txtView mcp.jsonHelp articles

10

CRM tools

3

Resources

5

Prompt templates

Org scoped

Data boundary

Tool catalog

What AI clients can do through MCP

Tools are grouped around contacts, notes, messages, and pipeline work. Read actions should be used first. Write, send, and delete actions should be gated by user confirmation inside the AI client.

Contactssearch_contacts

Search CRM contacts by name, company, email, phone, city, pipeline stage, status, priority, or source.

Safe use: Use before opening or updating a record. Results are scoped to the authenticated user's organization.

Contactsget_contact_details

Open the full contact profile, including core fields, address, social links, pipeline state, notes, and recent messages.

Safe use: Use when the AI needs a complete customer context before drafting a reply, summary, or next action.

Contactsupdate_contact

Update contact fields such as name, company, email, phone, WhatsApp, address, status, priority, stage, value, tags, notes, and next follow-up date.

Safe use: Use only after the user confirms the intended change. Call get_pipeline_stages first before changing stage values.

Contactsdelete_contact

Permanently delete a contact and its related notes and messages.

Safe use: Destructive action. The AI client should always ask for explicit confirmation before calling this tool.

Contactsget_customer_timeline

Build a chronological view of notes and messages for a contact.

Safe use: Use for account reviews, handoff preparation, sales call preparation, and follow-up planning.

Notescreate_note

Create a note on a contact for meeting summaries, observations, support context, or follow-up actions.

Safe use: Use after meetings, calls, inbox replies, or AI-generated summaries that the user wants saved into CRM history.

Notessearch_notes

Search organization notes by text.

Safe use: Use when the AI needs historical context that may not be in the latest contact fields.

Messagingsend_whatsapp

Send a WhatsApp message to a CRM contact using the organization's configured WhatsApp provider.

Safe use: External communication. The AI client should confirm the final message with the user before sending.

Messagingsend_email

Send an email to a CRM contact through the CRM email flow.

Safe use: External communication. The AI client should confirm subject and body before sending.

Pipelineget_pipeline_stages

List valid pipeline stages with keys, labels, definitions, order, and follow-up cadence.

Safe use: Use before updating pipelineStage so the AI writes valid stage keys instead of free text.

Resources

Resources exposed by MCP

crm://schema

crm-schema

Human-readable CRM schema reference for contacts, notes, messages, workflow triggers, and NFC profile fields.

crm://pipeline/{userId}

pipeline-definition

Pipeline stage definitions and cadence guidance for a specific authenticated user context.

crm://contact/{userId}/{contactId}

contact-summary

Compact contact summary with recent notes, recent messages, pipeline status, priority, score, and next follow-up.

Prompts

Prompt templates for repeatable work

sales_assistant

Reviews a contact relationship, identifies the next action, drafts follow-up, and flags risks or opportunities.

follow_up_drafter

Drafts a short personalized follow-up message using the last interaction, goal, and requested tone.

meeting_summary

Turns raw meeting notes into a structured summary, discussion points, decisions, action items, and next steps.

opportunity_risk_analysis

Analyzes a deal for risk level, risk factors, immediate action, timeline, and escalation recommendation.

support_response_drafter

Drafts a customer support response using the issue, interaction history, and proposed resolution.

Setup

Recommended implementation path

  1. 1Confirm that your PageCRM plan includes MCP server access.
  2. 2Ask the platform or organization admin to create an MCP client token from the admin AI settings area.
  3. 3Use the MCP endpoint https://pagecrm.com/mcp when the remote MCP gateway is enabled for the deployment.
  4. 4For self-hosted or internal deployments, use the configured MCP server endpoint, usually http://localhost:3001/mcp behind a secure proxy.
  5. 5Add the Authorization header as Bearer <your_mcp_token> in the MCP client configuration.
  6. 6Start with read-only actions such as search_contacts, get_contact_details, get_customer_timeline, and get_pipeline_stages.
  7. 7Enable write or send workflows only after confirming governance, audit expectations, and user approval steps.

Supported clients

Where MCP can be used

  • Claude Desktop
  • Claude for Teams or Enterprise clients that support remote MCP
  • Cursor and AI coding clients that support MCP over HTTP
  • Internal AI agents and automation runners
  • Custom business copilots that can send bearer-authenticated HTTP requests

Security

Governance rules for safe AI access

Every MCP request requires a bearer token from the MCP client registry.
Tokens are stored hashed in the application database and are not shown again after creation.
CRM data access is scoped to the authenticated user's organization.
Contact reads, timeline reads, notes, messages, and updates use organization-aware lookups.
External sending tools for WhatsApp and email should be treated as confirmation-required actions.
Destructive tools such as delete_contact should require explicit user confirmation in the AI client.
MCP clients can be disabled from the admin AI settings area if a token is no longer trusted.

FAQ

PageCRM MCP questions

What is the PageCRM MCP server?

The PageCRM MCP server lets approved AI clients use controlled CRM tools, resources, and prompts. It helps AI assistants search contacts, read timelines, create notes, update records, and prepare or send messages through governed workflows.

Can public AI systems access private CRM data?

No. The public MCP documentation is crawlable, but private CRM data requires an authenticated MCP bearer token and organization-scoped access.

Which MCP tools are available?

The current MCP server exposes contact search, contact details, contact updates, contact deletion, customer timeline, note creation, note search, WhatsApp sending, email sending, and pipeline stage lookup.

Can the MCP server send WhatsApp and email?

Yes, if the organization has configured the required communication channel. AI clients should always confirm final message content with the user before using send_whatsapp or send_email.

Is the MCP server organization-wise for SaaS customers?

Yes. MCP data access is scoped to the authenticated user's organization so one organization's contacts and activity should not be exposed to another organization.

Where should AI crawlers read PageCRM MCP documentation?

AI crawlers should read the public landing page at /mcp-server, the machine-readable text document at /mcp.txt, the structured manifest at /mcp.json, and the AI discovery files at /llms.txt and /llms-full.txt.