What is MCP?
The Model Context Protocol (MCP) is an integration layer that connects AI assistants directly to your Aloware account. Once connected, an AI assistant can execute real actions on your behalf using natural language. There is no need to switch between tools or write API calls manually.
You can instruct your AI client to send messages, make calls, create leads, enroll contacts into sequences, or manage power dialer lists, and the action is performed inside Aloware in real time.
Example: A prompt like "Send an SMS to +18185551234 confirming the appointment tomorrow at 2pm" is executed immediately through Aloware. Credits are consumed the same way they would be for a manual or API-triggered action.
Where to find MCP in Aloware
To access MCP, go to Aloware admin → Integrations → MCP.
This page includes everything needed to connect your AI client, including:
Your API token
MCP server endpoints
Ready-to-copy configuration for supported clients
Authentication and access
Aloware MCP supports two (2) authentication methods. Choose the one that matches how you are connecting.
OAuth 2.1 (recommended)
OAuth 2.1 is used for marketplace integrations such as the Claude Connectors Directory and ChatGPT Apps directory. It uses a standard consent flow, where the user approves access through their AI client, and supports user-level identity, rate limiting, and audit tracking. This method is required for marketplace listings and enterprise-grade integrations.
When to use it: You are connecting Aloware through Claude AI (web) or ChatGPT's built-in connector directory.
Bearer Token
Bearer Token authentication is used for direct client configuration, typically desktop AI clients that accept JSON config files. It uses the same Form Capture API token found in your MCP integrations page, which is manually added to the AI client configuration file.
When to use it: You are connecting Claude Desktop, Cursor, Windsurf, or another MCP-compatible client that does not support OAuth.
How to configure MCP using a bearer token
When using Bearer Token authentication, configuration is done through your AI client’s MCP config file. The setup registers each MCP server and includes your API token in the request headers.
{
"mcpServers": {
"aloware-communication": {
"url": "https://app.aloware.io/mcp/communication",
"headers": {
"Authorization": "Bearer [API_TOKEN]"
}
},
"aloware-contact": {
"url": "https://app.aloware.io/mcp/contact",
"headers": {
"Authorization": "Bearer [API_TOKEN]"
}
},
"aloware-operations": {
"url": "https://app.aloware.io/mcp/operations",
"headers": {
"Authorization": "Bearer [API_TOKEN]"
}
}
}
}Replace [API_TOKEN] withe the token found in your Aloware MCP integrations page. Once saved, the client will use the token to authenticate every request sent through MCP.
How MCP works
MCP connects your AI client to Aloware through streaming servers. Unlike traditional APIs that use request-and-response, MCP uses persistent HTTP Streaming (SSE). The AI client maintains a continuous connection to Aloware and executes actions in real time during your conversation.
All actions triggered through MCP behave the same as actions performed via the Aloware UI or API. This applies to billing, rate limits, and logging.
MCP servers and capabilities
Aloware MCP is divided into three servers, each handling a specific type of action. You can connect to one or all three, depending on your use case.
Connecting all three provides full MCP functionality.
Communication server
Endpoint: /mcp/communication (HTTP Streaming)
Purpose: Sending messages and making calls
The Communication server is used when the AI assistant needs to reach out to a contact through Aloware. It handles outbound communication actions such as sending SMS, leaving ringless voicemails, and placing calls.
For example, if a user tells their AI assistant to send a confirmation message, call a lead, or leave a voicemail, the request is executed through this server inside Aloware.
Supported operations
Operation | What it does | Paid |
Send SMS/MMS | Sends a text or MMS to a phone number using an Aloware line. Creates a new contact if the recipient doesn't exist. | Yes, credits per segment |
Send RVM | Sends a ringless voicemail using an audio file. Requires RVM to be enabled on the account. | Yes, credits per RVM |
Establish Two-Legged Call | Connects a contact and an agent (or ring group) in a live bridged call. | Yes, credits per minute |
Make AloAi Outbound Call | Triggers an outbound call using an AloAi voice agent. Requires an active voice bot configured for outbound calling. | Yes, credits per minute |
All actions in this server are real operations and consume credits based on usage. There is no simulation or test mode.
Required request details
All phone numbers must be in E.164 format (e.g. +18185551234), and each request must include either a from phone number or a line_id to identify which Aloware line to use.
If the recipient phone number does not exist in the account, a new contact is automatically created before the communication is executed.
Use cases
Appointment confirmation
"Text John Smith confirming his demo tomorrow at 3pm."
The AI sends an SMS through the assigned Aloware line.
Follow-up voicemail
"Leave a voicemail for everyone who didn't answer today's outbound list."
The AI issues RVMs to the specified contacts.
Warm transfer
"Call the contact at +18185551234 and bridge them to the sales ring group."
The AI establishes a two-legged call between the contact and an available agent.
AI-driven outreach
"Have the AloAi sales bot call the new leads from this morning."
The AI triggers AloAi outbound calls to the target contacts.
Tracking and billing
All actions are logged in the contact timeline and labeled with Creator Type: MCP, so MCP activity can be tracked separately from manual or API actions. These operations are real and consume credits. There is no sandbox or dry-run mode.
Contact server
Endpoint: /mcp/contact (HTTP Streaming)
Purpose: Managing contacts and leads
The Contact server is used when the AI assistant needs to create, update, or manage contact records inside Aloware. It also connects contacts to workflows such as sequences, tags, and power dialer lists.
Supported operations
Operation | What it does | Paid |
Create Lead | Creates or updates a contact. Supports standard and custom fields, user assignment, sequence enrollment, tags, Power Dialer, and ring group distribution, all in one request. | No |
Contact Lookup | Retrieves a contact by phone number. Returns the full contact record, including tags and custom attributes. | No |
Enroll Contact to Sequence | Adds a contact to a sequence or workflow. Supports lookup by Aloware ID, phone number, or CRM ID. | No |
Disenroll Contact from Sequence | Removes a contact from all active sequences using the same lookup options. | No |
Number Lookup (LRN) | Returns carrier, line type (mobile/landline/VoIP), and location data for a phone number. | Yes, credits per lookup |
Request requirements
Contact identification must use at least one supported method: phone number, Aloware contact ID, or CRM ID (HubSpot, Zoho, Guesty, or Pipedrive). When a phone number is used, it must follow E.164 format.
Use cases
Lead capture from a conversation
"Create a lead for Jane Smith, phone +18185551234, email [email protected], tag her as 'Event Signup', assign her to Mark, and add her to the Q2 Outreach sequence."
One request handles all of it.
Contact enrichment
"Look up the contact at +18185551234."
The AI returns the full contact record so the user can reference tags, notes, and custom fields.
CRM-driven enrollment
"Enroll HubSpot contact 12345 into the Renewal sequence."
The AI resolves the HubSpot ID and enrolls the matching Aloware contact.
Pre-call validation
"Run an LRN lookup on this list of numbers."
The AI identifies which are mobile vs. landline vs. VoIP before sending SMS.
Sequence cleanup
"Remove +18185551234 from all active workflows."
The AI disenrolls the contact immediately.
All MCP-created contacts are labeled with the intake source value mcp.
Tracking and billing
All contacts created through MCP are recorded with an intake source of mcp. Most operations are free. Only Number Lookup (LRN) consumes credits.
Operations server
Endpoint: /mcp/operations
Purpose: Team and workflow management
The Operations server gives the AI assistant visibility into users, agent status, and power dialer lists. It supports operational decisions like routing and list management but does not directly communicate with contacts.
Supported operations
Operation | What it does | Paid |
Get users | Lists all users in the account with their current agent status (Available, Busy, On Break, On Call, Wrap-up, Ringing, Offline). | No |
Inbox availability | Checks which agents are available for a specific ring group or inbox. Returns routing results explaining why each agent can or cannot receive calls (operating hours, concurrency, area code coverage, etc.). | No |
Remove contact from power dialer | Removes a specific contact from all power dialer lists they appear in. | No |
Clear power dialer list | Removes all contacts from a specific power dialer list. The list itself is preserved. | No |
Clear power dialer user lists | Clears all power dialer lists assigned to a specific user or agent. | No |
Use cases
Smart routing check
"Who is available on the Support inbox right now?"
The AI returns live agent availability before a supervisor assigns a call.
Routing diagnostic
"Why isn't Maria receiving calls in the Sales inbox?"
The AI runs an Inbox Availability check and returns routing reasons (e.g. outside operating hours, at concurrency limit).
Power dialer cleanup
"Clear the 'April Outreach' Power Dialer list."
The AI empties the list while keeping its configuration intact.
Agent reset: "Clear all Power Dialer lists assigned to John." → The AI removes all lists for that user in one action.
Remove a single contact
"Take +18185551234 out of all Power Dialer lists."
The AI removes the contact from every list it appears in.
Tracking and billing
All operations in this server are free. Power dialer operations are processed asynchronously. The system confirms the request as soon as it is accepted.
Billing summary
MCP operations that consume credits are the same operations that consume credits when performed through the UI or API.
Operation | Credit cost |
SMS | Per segment (160 characters each) |
MMS | Per message |
RVM | Per voicemail |
Two-legged call | Per call minute |
AloAi outbound call | Per call minute |
Number/LRN lookup | Per lookup |
Non-communication operations (creating contacts, enrolling in sequences, managing power dialer, checking availability) are free.
Important: AI assistants can execute paid operations autonomously based on your prompts. There is no sandbox mode. The MCP Integrations page includes a billing warning to make sure users understand this before connecting.
Tracking MCP activity
All communications created through MCP are tagged with Creator Type: MCP. These messages and calls appear in the contact timeline like any other communication, but they can be filtered and reported on separately from manual, API, or workflow-originated communications.
Contacts created via MCP are recorded with an intake source of mcp, which appears in the contact's intake source field.
Supported AI clients
MCP is an open standard, so any MCP-compatible client can connect to Aloware's MCP servers.
The most common clients are Claude Desktop (Anthropic) and Cursor, both with full MCP support, along with Windsurf. Claude AI (web) and ChatGPT also connect through their OAuth 2.1 marketplace connectors, and any other client implementing the MCP specification can connect as well.
The in-app MCP Integrations page provides ready-to-copy JSON configuration for Claude Desktop and Cursor.