WhatsMCP: MCP for WhatsApp

Start an AI Voice-Agent Call

wa_ai_call_start

Have an AI voice agent call a WhatsApp number from one of your numbers, using a preset from wa_ai_list_presets. Pass per-call context in variables. Returns immediately with a call_ref; the call then rings and runs on its own. Poll wa_ai_call_get about every 30 seconds until status is done, failed, no_answer or busy — then it carries the summary, transcript, collected data and a recording link. One call at a time per number; a number that just finished a call must wait a moment.

Input Schema

Name	Required	Description
`to`	Yes	the WhatsApp number to call, international format, e.g. +447700900123
`preset`	Yes	the preset name, from wa_ai_list_presets
`variables`	No	per-call context for the agent: flat string/number/boolean values, merged over the preset's defaults, e.g. {"customer_name":"Ann","order_id":"A-1042"}
`first_message`	No	override what the agent says first on this call

Output Schema

Name	Required	Description
`next`	No	what to do next
`status`	No	dialing once accepted
`refusal`	No	present only when the call was not placed
`call_ref`	No	pass to wa_ai_call_get to follow the call

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedOct 26, 2026

addedOutput schema / properties / refusal / properties / candidates

Added value: +{
+  "description": "on an account_ambiguous refusal: the accounts to choose from — retry with one of these as account_id",
+  "items": {
+    "additionalProperties": false,
+    "properties": {
+      "account_id": {
+        "description": "the id to retry with as account_id",
+        "type": "string"
+      },
+      "name": {
+        "description": "the account's WhatsApp display name",
+        "type": "string"
+      },
+      "phone": {
+        "description": "the account's phone number, E.164",
+        "type": "string"
+      },
+      "status": {
+        "description": "whether the number works right now — live, connecting, or why not, as wa_list_accounts reports it",
+        "type": "string"
+      }
+    },
+    "required": [
+      "account_id"
+    ],
+    "type": "object"
+  },
+  "type": [
+    "null",
+    "array"
+  ]
+}

Changed3 schema fields changedOct 11, 2026

addedOutput schema / properties / refusal / properties / account_status

Added value: +{
+  "description": "when the number is not working: its status — starting, not_running, whatsapp_down, suspended, pairing, unlinked or locked",
+  "type": "string"
+}

addedOutput schema / properties / refusal / properties / account_status_reason

Added value: +{
+  "description": "the detail of account_status, e.g. no_egress, stopped, no_plan",
+  "type": "string"
+}

addedOutput schema / properties / refusal / properties / since

Added value: +{
+  "description": "when the number stopped working, RFC3339, when known",
+  "type": "string"
+}

AddedSep 26, 2026

TDQS

A4.7/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Goes well beyond the annotations: it discloses that the tool is asynchronous ('returns immediately with a call_ref; the call then rings and runs on its own'), names the terminal statuses, and states the concurrency/cooldown limits that explain the non-idempotent, non-read-only annotation profile. This is the kind of context the structured fields cannot carry.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Four sentences, front-loaded with the action, then the return shape, then the polling loop, then the constraints. No filler and nothing important is buried.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For an async mutation tool with an output schema and full schema coverage, the description supplies everything an agent needs: how the call is started, that it returns a call_ref, how to observe completion, and the concurrency limits that cause failures.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so every parameter is already documented in the schema, including variables merge semantics and format examples. The description only restates that per-call context goes in variables and the preset comes from wa_ai_list_presets, adding little beyond the schema baseline.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Have an AI voice agent call a WhatsApp number from one of your numbers') plus the required configuration source (a preset). It is unambiguously distinct from wa_ai_call_get, wa_ai_call_hangup and wa_list_calls, which it does not rely on the name alone to separate.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicit about the prerequisite (preset from wa_ai_list_presets), the follow-up workflow (poll wa_ai_call_get about every 30 seconds until status is done/failed/no_answer/busy), and the constraints (one call at a time per number, cooldown after a call finishes). An agent knows exactly when to use this and what to do next.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All WhatsMCP: MCP for WhatsApp tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?