# SigmaMind AI MCP > SigmaMind AI connects voice workflows to your agent, letting you manage calls and analyze conversations through your preferred AI client. You can initiate a call using `create_call` or list all existing ones with `list_calls`. The server exposes tools for full transcript retrieval (`get_call_transcript`), sentiment analysis (`get_call_analysis`), and configuring dedicated voice agents via `create_agent`. It's built for automating communication processes without writing complex backend code. ## Overview - **Category:** communication-messaging - **Price:** Free - **Tags:** voice-agents, call-automation, sentiment-analysis, transcription, conversational-ai, call-management ## Description Listen up. This SigmaMind AI MCP Server connects your agent directly to a whole stack of voice communication tools through **your AI client**. You don't gotta write complex backend code; you just point your agent at these functions and it handles the heavy lifting. You need to know what numbers you got? Use `list_numbers` to display every phone number available for calling campaigns. Before you run anything, check the connection health first. Running `check_sigmamind_status` verifies if the API link to SigmaMind is live and working right. When it's time to build out your communication stack, you can start by checking what AI agents are already set up with `list_agents`. If you need a new personality for specific tasks, you use `create_agent` to generate a custom-prompted voice agent profile. Once that’s done, if you want details on the agent you just made or one from before, run `get_agent` by its ID to pull up all the info. Ready to make a call? You initiate a live conversation using `create_call`, specifying both the target number and the designated voice agent. When that call wraps up, you can grab basic metadata for the session—whether it’s still going or finished—by calling `get_call`. But getting the basics isn't enough. You gotta understand what was said. Use `get_call_transcript` to retrieve the full text transcript of any completed voice call, complete with time stamps so you know exactly when things were mentioned. For deep analysis, running `get_call_analysis` gives you structured insights: it pulls out sentiment scores and breaks down key topics from that recorded conversation based on a specific call ID. To keep track of the calls you run—finished or still ongoing—you can pull up a list of recent sessions using `list_calls`, which provides IDs and status for quick tracking. If you've got an agent ID, running `get_agent` lets you retrieve all the detailed information about that specific voice personality. Basically, your workflow runs like this: You check the system health with **`check_sigmamind_status`**. Then, you manage who’s talking by using **`list_agents`** and setting up a custom persona with **`create_agent`**. When you're ready to go out, you trigger it all with **`create_call`**, pointing it at a number retrieved from **`list_numbers`**. Once the hang-up happens, your agent can pull the full text using **`get_call_transcript`** and generate concrete performance metrics—sentiment scores, key topics, etc.—by executing **`get_call_analysis`** on that call ID. You can also check basic details of any session with **`get_call`**, or look up specific agent info again with **`get_agent`**. It's a full loop for managing and analyzing every single voice conversation. ## Tools ### check_sigmamind_status Verifies if the API connection to SigmaMind is active and functioning correctly. ### create_agent Creates a new, custom-prompted AI voice agent profile for specific communication tasks. ### create_call Initiates a live voice call to a specified number using a designated agent. ### get_agent Retrieves detailed information about a specific AI agent by its ID. ### get_call_analysis Generates structured insights, including sentiment scores and topic breakdowns, from a recorded call. ### get_call Gets basic details and metadata for a completed or ongoing call session. ### get_call_transcript Retrieves the full, time-stamped text transcript of any completed voice call. ### list_agents Lists all existing AI agents configured in your SigmaMind account. ### list_calls Retrieves a list of recent calls, providing IDs and status for easy tracking. ### list_numbers Displays all associated phone numbers available for calling campaigns. ## Prompt Examples **Prompt:** ``` List all my AI voice agents. ``` **Response:** ``` You have 3 agents: 'Sales Qualifier' (42 calls), 'Support Bot' (128 calls), 'Survey Agent' (15 calls). ``` **Prompt:** ``` Call +14155551234 with agent 'Sales Qualifier'. ``` **Response:** ``` Call initiated! ID: call_8291 to +14155551234 using 'Sales Qualifier'. The AI agent is now handling the conversation. ``` **Prompt:** ``` Show transcript for call call_8291. ``` **Response:** ``` Transcript for call_8291 (3m 42s): Agent greeted the contact, qualified interest in the product, and scheduled a demo for next Tuesday. ``` ## Capabilities ### Manage AI Agents List existing agents with `list_agents`, or create new voice personalities using `create_agent`. ### Execute and Track Calls Initiate a live, recorded conversation via `create_call`, and retrieve the status of any call through `get_call`. ### Extract Conversation Details Pull full text transcripts from completed interactions using `get_call_transcript`. ### Analyze Call Outcomes Run `get_call_analysis` to receive machine-generated data on sentiment, key topics, and performance metrics for a call ID. ### View System Status Check the API connection health using `check_sigmamind_status` before running any workflow. ## Use Cases ### Lead Qualification Audit A sales rep needs to audit a week's worth of calls. Instead of downloading 20 separate recordings, they ask their agent: 'Get the analysis for all my calls last Tuesday.' The agent runs `list_calls`, then loops through IDs, calling `get_call_analysis` on each one. They end up with a single JSON object containing sentiment scores and topic breakdowns for every call. ### Support Bot Deployment The support team is rolling out a new automated bot. Before deployment, they use `list_agents` to check the current setup. They then run `create_agent`, uploading a detailed script and prompt. Finally, they test it by calling `create_call` with a dummy number, validating that the agent performs as expected. ### Debugging Call Flow A call fails midway through. Instead of guessing what went wrong, an engineer uses `get_call` to check the basic metadata for the failing ID. If the data looks right, they run `get_call_transcript` and immediately see where the conversation dropped or deviated from the expected script. ### Multi-Campaign Reporting A manager runs three different campaigns (Sales, Survey, Support). They use `list_numbers` to verify all associated lines. Then they ask the agent to summarize all call data using a sequence of calls: `create_call`, followed by gathering transcripts via `get_call_transcript` for each run. ## Benefits - **Analyze Outcomes Instantly:** Stop reviewing PDFs. Use `get_call_analysis` to get instant, structured data on sentiment (positive/negative) and key topics from any call ID. This turns raw audio into quantifiable metrics. - **Manage Agents via Code:** Don't mess with web dashboards. Your agent uses `list_agents` and `create_agent` to deploy multiple voice personas—one for sales, one for support, etc.—all controlled by simple prompts. - **Full Audit Trail on Demand:** Need to see exactly what was said? Call `get_call_transcript`. It pulls the complete text log of a call, making it trivial to search for specific keywords or quotes later. - **Automate Campaign Execution:** Start calls with one command. Calling `create_call` lets your agent handle complex conversations (like qualifying leads) without any manual intervention from you during the process. - **Centralized Call Oversight:** Get a birds-eye view of everything using `list_calls`. You track call history, check status via `get_call`, and maintain accountability across large teams. ## How It Works The bottom line is that your AI client manages the entire voice lifecycle—from setup through execution and analysis—without you needing to interact with a web dashboard. 1. First, use `list_agents` to confirm existing agent profiles. If you need a new one, call `create_agent`, providing the necessary prompt context. 2. Next, call `create_call`, passing the target phone number and the specific agent ID. This starts the live conversation flow. 3. Finally, once the call concludes, execute `get_call_analysis` with the resulting call ID to pull structured sentiment scores and topic breakdowns. ## Frequently Asked Questions **How do I find out if I can run a call with SigmaMind AI using `create_call`?** You must first check the status and availability. Run `check_sigmamind_status` to confirm API connectivity, then use `list_numbers` to verify an active number is assigned before calling `create_call`. **What data do I need for `get_call_analysis`?** You only need the call ID. The tool runs the analysis on the recorded session associated with that ID, returning sentiment and topic breakdowns. **Can I reuse an agent created with `create_agent` for a different campaign?** Yes. Once you run `get_agent`, you'll see its details and prompts; those configurations can be reused or slightly modified to handle new campaigns via subsequent calls. **What is the difference between `list_calls` and `get_call`?** `list_calls` gives you a list of call IDs and basic status flags. `get_call` takes one specific ID and retrieves detailed metadata about that single, specified call. **How does `check_sigmamind_status` verify my API connectivity?** It confirms your API connection is active and ready for use. The tool returns a success code or details the specific error, letting you know immediately if there's an issue with authentication or credentials before you start running calls. **What happens if I call `create_call` but the conversation fails?** The system records a detailed failure status that you can check using `get_call`. This lets you retrieve error codes and know exactly why the call dropped or couldn't connect, saving you time on troubleshooting. **Are there any usage limits or rate limits for calling tools like `create_call`?** SigmaMind enforces standard API rate limiting to ensure system stability. If you hit a limit, the tool will return a 429 error code, telling your AI client exactly when you can try again. **How do I handle large amounts of data from `get_call_transcript`?** The API handles transcripts up to a specified maximum length. For extremely long calls, the tool provides pagination metadata, allowing your agent to fetch the transcript in manageable chunks rather than one massive block of text. **Can my AI initiate phone calls?** Yes. Use `create_call` with an agent ID and phone number. The AI agent handles the conversation automatically. **Can I read call transcripts?** Yes. `get_call_transcript` returns the full conversation for any completed call. **How do I create a new voice agent?** Use `create_agent` with a name and system prompt. The agent will use the prompt to guide conversations.