Protocols

Protocol-agnostic scheduling

Temporal Cortex is not just an MCP server — it’s scheduling infrastructure accessible via 4 protocols. The same tools, the same computation, the same safety guarantees, available through whichever protocol fits your architecture.

MCP (Model Context Protocol)

Pattern: Agent ↔ Tools

The primary integration path for AI agents. Your agent calls scheduling tools directly via the Model Context Protocol.

Best for: Claude Desktop, VS Code/Cursor/Windsurf, any MCP-compatible agent framework.

# Install and run
npx @temporal-cortex/cortex-mcp

# Supports both transports
TRANSPORT=stdio npx @temporal-cortex/cortex-mcp    # Default
TRANSPORT=http npx @temporal-cortex/cortex-mcp     # Streamable HTTP

Supported frameworks: Claude Desktop, Cursor, Windsurf, VS Code, Zed, LangGraph, CrewAI, OpenAI Agents SDK, and 20+ more via Agent Skills.

A2A (Agent-to-Agent)

Pattern: Agent ↔ Agent

Agents negotiate scheduling with other agents directly over Google’s open A2A protocol. Each user publishes an Agent Card declaring their scheduling capabilities. Discovery happens via resolve_contact → Agent Card URL.

Best for: Cross-organization scheduling where both parties have AI agents.

How it works:

  1. Your agent calls resolve_contact to discover the other person’s scheduling path
  2. If an Agent Card is found → negotiate directly via A2A
  3. If no Agent Card → fall back to compose_proposal for email-based scheduling

This is protocol negotiation — the agent discovers the best path automatically.

REST API

Pattern: App ↔ API

Standard HTTP endpoints for embedding scheduling directly into your product. Build custom UIs, integrate with existing backends, or orchestrate from server-side code.

Best for: SaaS products, custom scheduling UIs, backend integrations.

Base URL: https://api.temporal-cortex.com (Platform Mode)

Authentication: API key in Authorization: Bearer sk_live_... header.

Pattern: Human ↔ Booking

Public booking pages for when the other person doesn’t have an agent. Share a link, they pick a time, the booking is atomic.

Best for: External-party scheduling, backward compatibility, “schedule with anyone” workflows.

How it works:

  1. Enable Open Scheduling in the platform dashboard
  2. Get your Temporal Link: book.temporal-cortex.com/your-handle
  3. Share the link — recipients pick available times
  4. Booking uses the same Two-Phase Commit as agent-driven booking

Protocol negotiation workflow

The agent discovers the best scheduling path based on the other person’s capabilities:

resolve_contact(email: "[email protected]")
├── Agent Card found → A2A negotiation (fastest)
├── Temporal Link found → share booking link
├── Email only → compose_proposal → email proposal
└── Phone only → compose_proposal → SMS proposal

This backward-compatible approach means your agent can schedule with anyone — whether they have an AI agent, a booking page, or just an email address.