Let your AI agent place real phone calls. A real person makes the call and returns structured answers plus wrap-up notes
TaskBullet's MCP server lets an AI agent place a real phone call that a real person makes. The agent supplies the phone number, goal and questions; a TaskBullet assistant makes the call and returns structured answers matched to each question plus wrap-up notes. Tools: validate_call, place_call, get_call, update_call, cancel_call, get_transcript, plus delegate_task for other jobs an agent can't finish. US and Canada numbers. Pricing: $2.50 for the first minute (minimum), then $1 per additional minute. One MCP connection reaches real tools, and a real person finishes the job when AI gets stuck. Remote endpoint: https://taskbullet.com/api/hub/mcp (Streamable HTTP); also stdio via npm. Website: https://taskbullet.com/hub. Contact: [email protected]. Author: TaskBullet.
from agents import Agent
from agents.mcp import MCPServerStdio
mcp_server = MCPServerStdio(
command="npx",
args=["-y", "@taskbullet/mcp-server"],
)
agent = Agent(
name="My Agent",
model="gpt-4o",
mcp_servers=[mcp_server],
)If you maintain TaskBullet (Human Phone Calls + task handoff), add this badge to your README to show it's verified on CuratedMCP:
[](https://www.curatedmcp.com/marketplace/taskbullet-human-phone-calls-task-handoff)
Memory MCP Server
Persistent memory for AI agents using a local knowledge graph β remember facts across sessions
Sequential Thinking MCP
Structured step-by-step reasoning for complex problem solving and multi-step analysis
Caisey
Troubleshoot enrolled client PCs from your AI client, with human approval and an audit trail.