Vapi MCP. Automate voice calls and extract call transcripts.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Vapi lets your AI agent handle automated voice calls directly from your chat. You can trigger outbound phone calls (PSTN/WebRTC), update complex assistant personas on the fly, and pull full call transcripts instantly.
It turns your standard conversational workspace into a live telephony dispatch center without ever leaving your IDE.
What your AI agents can do
Create assistant
Builds a new Voice AI assistant persona by providing configuration details like the model and voice ID.
Create phone call
Starts an outbound phone call immediately, requiring only the target number ID and customer info.
Create web call
Generates a unique link for setting up a web-based voice call session.
Initiate outbound calls via create_phone_call or generate web links using create_web_call, and pull all associated data with get_call_details.
Create new assistant personas, adjust their underlying models, or update system instructions instantly using create_assistant or update_assistant.
See which numbers are active (list_phone_numbers), list all configured assistants (list_assistants), and map out complex agent teams (list_squads).
List recent activity with list_calls to find transcripts or details for any past conversation.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Vapi MCP Server: 10 Tools for Call & Voice Ops
Manage all aspects of automated voice communication, from creating new assistant personas to retrieving detailed transcripts for any call.
019d761bcreate assistant
Builds a new Voice AI assistant persona by providing configuration details like the model and voice ID.
019d761bcreate phone call
Starts an outbound phone call immediately, requiring only the target number ID and customer info.
019d761bcreate web call
Generates a unique link for setting up a web-based voice call session.
019d761bget call details
Pulls detailed records, full transcripts, and metrics for any specific recorded phone call ID.
019d761blist agent tools
Lists every available tool that the Vapi assistants can use when running a conversation.
019d761blist assistants
Retrieves a list of all currently configured Voice AI assistant personas in your account.
019d761blist calls
Gets a summary view of both recent and active voice calls managed by Vapi.
019d761blist phone numbers
Displays all phone numbers currently connected to your Vapi account for calling purposes.
019d761blist squads
Lists the available multi-agent 'squads' used for configuring advanced, multi-step conversational paths.
019d761bupdate assistant
Modifies any existing assistant’s configuration, like changing its core prompt or model version.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Vapi, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
Your AI agent handles automated voice calls right from your chat interface. You'll never have to leave your IDE just because you need to talk to someone or track a conversation.
Managing the Entire Call Lifecycle
You can kick off phone communications instantly using create_phone_call, which starts an outbound call to a target number ID, handling both PSTN and WebRTC connections. For web-based interactions, you just run create_web_call and it generates a unique link for setting up the voice session.
When the call's done—or even if it’s still happening—you pull all the data with get_call_details. This single function gives you full transcripts, detailed metrics, and records for any specific phone call ID. To keep track of what’s been going down lately, use list_calls to get a summary view of both recent activity and currently active voice calls.
Defining and Customizing Your AI Personas
Building out the intelligence is simple. You create new assistant personas using create_assistant, passing in everything from the model name to the specific voice ID you want it to use. If an existing assistant needs a refresh, you can modify its setup instantly with update_assistant. This lets you change things like core system prompts or switch out the underlying LLM model without breaking anything.
To see what personas you've already built, just run list_assistants to retrieve a list of all configured Voice AI assistants in your account. You can also check exactly which tools your assistant is allowed to use during a chat by calling list_agent_tools.
Auditing Resources and Complex Flows
You gotta know what numbers you're working with. list_phone_numbers displays every single phone number connected to Vapi, so you always know which line is active for calls. For setting up advanced conversations—the multi-step stuff—you can use list_squads. This function lists the available 'squads,' which are specialized groups used when configuring complex, multi-agent paths.
Finally, if you need a quick overview of your setup and what's available for calling, you run list_agent_tools to map out every toolset.
Every step is covered: initiating calls, getting the raw data back, customizing the AI brain, and managing all the underlying telecom resources. You get a full dispatch center right in your chat.
How Vapi MCP Works
- 1 Subscribe to the server and plug in your Vapi Private API Key.
- 2 Ask your AI agent to perform an action, like 'Call John Doe' (triggering
create_phone_call). - 3 The agent executes the tool call, manages the live voice connection, and returns structured data or a transcript snippet.
The bottom line is: Your AI client becomes a full-featured dispatch center for automated voice comms, triggered entirely through chat commands.
Who Is Vapi MCP For?
This is for the Ops Manager or Support Engineer who gets tired of manually checking separate CRM dashboards and telephony logs. If your job involves automating follow-ups, capturing call data, or setting up complex agent workflows, this tool saves you from context switching.
You use the server to tell your agent to run a demo outbound sales call to a prospect's number right inside your editor.
After an unhappy customer calls, you instantly pull the full voice transcript using get_call_details so you can summarize action items for follow-up tickets.
You need to iteratively test different LLMs or TTS voices. You use update_assistant to tweak the core parameters of an assistant without restarting your development environment.
What Changes When You Connect
- Stop switching tabs to grab call data. Use
get_call_detailsdirectly in your chat client to pull raw transcripts, metrics, and full call logs instantly. - Your agent handles the calling logic. Instead of writing complex phone scripts, just tell it: 'Call X' (
create_phone_call), and let it manage the PSTN connection. - Tweak agents without leaving your workflow. Use
update_assistantto change an LLM model or system prompt on a running persona in seconds. - Map complex interactions easily. Check out
list_squadsto configure multi-agent workflows, allowing one conversation to pass through several specialized bots. - Full visibility into resources. Run
list_phone_numbersto audit all connected lines and ensure the right number is assigned for a new campaign.
Real-World Use Cases
Need to qualify leads post-webinar
A sales engineer needs to follow up on 20 hot leads. Instead of manually calling them all, they ask their agent: 'Call the top five leads.' The agent uses create_phone_call repeatedly, and then uses get_call_details afterward to pull transcripts, summarizing which lines need manual intervention.
Summarize a bad customer call
A support manager receives a recording of an escalated complaint. They prompt their agent with the call ID: 'Summarize this issue.' The agent runs get_call_details, pulling the full transcript, and returns three clear action items for the support team.
Update an assistant's knowledge base
A product manager launches a new feature. They tell their agent: 'Update the Support Matrix bot with Feature X documentation.' The agent uses update_assistant, changing the system prompt and model, ensuring all future calls are accurate.
Test a multi-step conversation flow
You want to test if an initial lead screening must pass through a 'Data Collection' bot before reaching a 'Billing' bot. You run list_squads to configure the sequence, then use the agent to execute the call path.
The Tradeoffs
Trying to manually update model parameters.
Manually changing the model name in a configuration file or attempting to pass JSON directly into the chat prompt without structure.
→
Don't mess with config files. Use the update_assistant tool. Just tell your agent: 'Change the Support bot's model to GPT-4o-mini,' and it handles the syntax.
Forgetting which number to call.
Starting a campaign run and realizing halfway through that you used the wrong department phone number, wasting time and calls.
→
Always check list_phone_numbers first. Confirm the exact ID of the line before triggering any outbound activity with create_phone_call.
Assuming the agent remembers old call details.
Asking the chat agent, 'What did we talk about last week?' and getting a generic or incorrect answer because context was lost.
→
Always reference the ID. Use get_call_details with the specific call ID to force the retrieval of accurate transcripts.
When It Fits, When It Doesn't
Use Vapi if your core problem involves voice communication, needing to automate calls, or extracting structured data from phone conversations. If you only need simple text back-and-forth (like a basic FAQ bot), use a standard chat tool.
Don't use this if: 1) You are building an SMS service; those require messaging tools. 2) Your interaction is purely visual and never involves speaking to a human. Instead, focus on the 'Calling' capabilities. If you only need to see what bots exist without calling them, list_assistants works fine, but if you need to change something, use update_assistant.
In short: Voice + Data Retrieval = Vapi.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Vapi. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 10 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Manual call follow-up is a huge time sink.
Right now, if you need to follow up on a conversation—say, summarizing what happened during an initial qualification chat—you have to jump between three places: the CRM dashboard (to find the record), the call recording platform (to listen to the transcript), and then back to your notes app (to write out action items). It's tedious copying, pasting, and reading through raw audio.
With Vapi MCP Server, you just tell your agent the call ID. It runs `get_call_details`, pulls the full voice-to-text transcript directly into your chat window, summarizes it, and gives you clear action items—all without leaving your IDE.
Vapi MCP Server: Manage call routing & transcription
Before Vapi, setting up an automated campaign meant calling a specialized telephony platform API, managing credentials separately, and then hoping the resulting data could be manually fed back into your main workflow. It was siloed, slow, and required multiple developers just to bridge the gap.
Now, you treat it like any other tool call. You use `create_phone_call` directly from chat. The AI client handles the entire connection stack—the dialing, the conversation, the transcript—and feeds all the actionable data back into your single workspace.
Common Questions About Vapi MCP
How do I make an agent call a prospect using Vapi's create_phone_call tool? +
You tell your agent to initiate the call, providing the target phone number ID and customer details. The agent executes create_phone_call, establishing the outbound connection instantly.
What is the difference between list_calls and get_call_details? +
list_calls shows you a summary of recent or active calls, giving you an overview. You must use get_call_details with a specific call ID to retrieve the full transcript and metrics.
Can I change my assistant's model on the fly? Use Vapi's update_assistant tool. +
Yes. You use update_assistant to modify any existing persona, including changing its underlying LLM or system prompt without downtime.
How do I configure a multi-agent call flow with Vapi? +
You first run list_squads to see what's available. Then, you use the agent to set up and execute advanced conversational pathways using those defined squads.
What does the `list_agent_tools` tool reveal about my Vapi setup? +
It shows every specific function your AI client can execute through Vapi. This list confirms all available capabilities—like calling, updating personas, or listing squads—so you know exactly what actions your agent supports.
I need to find out which numbers I can use; how do I use `list_phone_numbers`? +
Running this tool gives you a complete roster of all connected phone lines. You get the necessary IDs and details from this list before sending those credentials to the create_phone_call tool.
When should I use the `create_assistant` tool? +
Use this when you need a brand-new AI persona. You define its core logic—including the model, voice settings, and transcriber type—all in one JSON payload before deployment.
How do I generate a link for an internal web meeting? Is it different from calling? +
Use create_web_call. This generates a specific web-based voice call link that you share with users. It bypasses traditional phone number routing and gives instant access to the conversation interface.
Can the agent place an outbound phone call for me? +
Yes! Through the createPhoneCallTool, you can ask your agent: Initiate a phone call to +15551234567 using our 'Sales Assistant' persona. It will format the payload correctly and instruct Vapi to dial the human using the specified AI persona.
How can I read the transcript of an old automated voice call? +
Just tell your chat assistant: fetch the transcript details for call ID '3fbea5x...'. The getCallTool reaches into Vapi, extracts the raw audio-to-text transcript logs, and the assistant can even summarize the entire human-to-AI interaction for you.
Can I tweak system prompts or models sequentially in my chat editor? +
Yes, especially if you're in an IDE like Cursor. You can iterate: Update Assistant 'Agent A' to use GPT-4o and set its prompt to 'Always be highly overly enthusiastic'. The agent will use the updateAssistantTool to reconfigure the live voice persona instantly.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Retell AI
Empower your conversational AI to orchestrate, analyze, and automate phone calls or web-based voice agent interactions via Retell.
Groq
Empower LLM applications via Groq — perform ultra-fast LPU-accelerated chat completions, handle audio transcription and translation, and use JSON mode directly from any AI agent.
AirOps
AI workflow orchestration — execute models, manage agents, and query memory via AI.
You might also like
Heap
Automate product analytics via Heap — track events, identify users, and manage segments directly from any AI agent.
Weather (Open-Meteo)
Get real-time weather, high-precision forecasts, air quality data, and severe weather alerts for any city worldwide.
Better Stack
Monitor uptime and incidents via Better Stack — list monitors, heartbeats, and on-call schedules directly from any AI agent.