What is ContactShip MCP?
Imagine managing all your voice agents just by talking to your favorite AI assistant. Instead of logging into the platform, finding the agent, editing the prompt, configuring the voice, saving, testing, and repeating… you simply say:“Create a collections agent for Mexico that sounds professional. Extract whether the customer committed to pay and the amount.”Review the proposed change and the tool result to confirm it was applied. That’s ContactShip MCP: a bridge that connects your ContactShip account with AI assistants like Claude, Cursor, or Claude Code. You talk in plain language, and the AI handles all the technical stuff.
What can you do?
Manage agents
View, create, and edit your voice agents. Change prompts, voice, language, speed — all from the chat.
Launch calls
Request an AI phone call, or schedule one for later.
Search contacts
Find contacts by name, email, or phone. View complete call history.
See your numbers
View your available phone numbers for outbound calls.
Review call results
Ask your AI to review all your voice agent call results at scale — find patterns, check quality, get summaries — without listening to recordings one by one.
How does it work?
1
Pick your AI tool
Works with Claude (web, desktop, and Code), Cursor, and any MCP-compatible app.
2
Connect your account
Paste your ContactShip API key and you’re done. Follow the client’s connection steps.
3
Talk to your AI
Ask for what you need in plain language. The AI understands the context and takes action for you.
Real examples
Edit an agent:“Make my sales agent sound warmer and change the language to Mexican Spanish.”Launch a call:
“Call Juan Pérez at +52 55 1234 5678 with the survey agent.”Search for a contact:
“Find the contact with email maria@company.com and show their call history.”Configure post-call analysis:
“Add a field to the collections agent to track whether the customer committed to pay, the amount, and the date.”Review call results:
“Show me all calls from today and summarize the results.”
“Get the transcript of the last call with Juan Pérez.”
“Which calls this week had a positive outcome? Give me a summary.”
“Find calls where the customer mentioned a payment issue.”
Reviewing call results at scale
With thelist_calls and get_call MCP tools, your AI assistant can access all your voice agent call results directly — no need to manually open recordings or read transcripts one by one.
Ask your AI to:
- List and filter calls — by date, agent, contact, or outcome
- Get full call details — transcript, extracted variables, duration, recording
- Find patterns — which calls succeeded, which had objections, what customers said most
Tools and publishing
Usesummarize_calls for aggregate results and search_calls to find mentions. list_calls returns compact rows; get_call reads one call in detail.
Voice agents support staging with update_agent_draft and explicit publication with publish_agent. update_agent publishes directly and may reject the request when a draft is open. Say whether you want a draft for review or a change applied to new calls.
The server also includes contact, board, task, and WhatsApp tools. Text-agent, conversation, and analytics tools depend on enabled services. Check the tools shown by your connected client to confirm availability.
Requirements
A ContactShip account with a plan that includes API access
Your API key — generate it from Settings > API Keys
A compatible AI assistant (Claude, Cursor, or any MCP-enabled client)
Each API key belongs to a single organization. If you manage several, see Multiple organizations.
Connect your tool
Claude (web)
Connect from claude.ai to manage agents right from your browser.
Claude Code
Connect from the terminal for development and automation workflows.
Cursor
Connect from Cursor IDE to manage agents without leaving your editor.
Multiple organizations
One connection per organization, each with its own API key.