Main idea:
What do I want this device or software to do, and what kind of connection or compatibility do I need to make that happen?
This quick-reference guide explains protocols — what they are, how they differ from ports and cables, and how to choose the right ones for what you're trying to do.
A protocol is a set of rules that define how devices or software communicate — whether it's sending video to a monitor, transferring files between systems, or securing a login. Some protocols operate entirely in software, while others rely on specific hardware to function.
When you're setting up an external monitor, you're really asking:
Can my laptop send video, and does this cable support video output?
→ You need a protocol like DisplayPort Alt Mode over USB-C or a classic HDMI connection.
When you're working with machine learning systems that use microservices (like a model served by an API), you may use gRPC — a high-performance protocol for remote procedure calls, especially common in Google Cloud, TensorFlow Serving, or internal ML workflows. Alternatively, HTTP/HTTPS or SFTP can handle large file downloads or remote jobs.
When you want to secure a login or API call, you're relying on protocols like OAuth or TLS — even though you might not see them.
All of these are powered by protocols — sets of rules that make devices and software talk to each other effectively.
| Use Case | Protocols |
|---|---|
| Web browsing | HTTP, HTTPS |
| Secure login/API auth | TLS, OAuth |
| Video output | HDMI, DisplayPort, USB-C (Alt Mode) |
| File transfer | FTP, SFTP, SCP, gRPC |
| Remote ML workflows | gRPC, HTTP, SFTP |
| Smart devices | Bluetooth, Zigbee, Z-Wave |
| Device control & sync | MIDI, CEC (for HDMI), USB protocols |
These protocols work below the surface — you don’t usually interact with them directly, but every higher-level protocol depends on them.
| Protocol | Layer | Role |
|---|---|---|
| IP (Internet Protocol) | Network | Routes packets between devices across networks |
| TCP (Transmission Control Protocol) | Transport | Ensures reliable, ordered data transfer |
| UDP (User Datagram Protocol) | Transport | Faster, less reliable alternative to TCP |
For example: HTTP relies on TCP, and TLS secures TCP connections.
They're all stacked together — this is called the network protocol stack.
These software-based protocols manage how AI systems (especially language models) interact with external tools, data, memory, and APIs in structured workflows.
| Protocol | Purpose |
|---|---|
| MCP (Model Context Protocol) | A standardized way for language models to access tools, files, memory, and APIs in structured, context-aware workflows. Used by systems like Claude to interact with real-world data. |
| OpenAI Tool Calling Schema | Defines message formats for GPT models to call tools and use the results. JSON-based and used in OpenAI’s assistant API. |
| LangChain Agent Frameworks | Organizes tool use, memory, and step-by-step reasoning loops for building agents. Works across many models and APIs. |
| Name | Owner | Open Standard? | Use Case |
|---|---|---|---|
| MCP | Hugging Face | ✅ Yes | Cross-model tool/memory interaction |
| OpenAI Tool Schema | OpenAI | ❌ No | GPT-specific tool calling |
| LangChain Agent Framework | LangChain | ❌ No | Custom agent building with memory/tools |
These protocols are key when building tool-using AI systems. MCP aims to standardize the space, while OpenAI and LangChain offer more framework-specific solutions.
Understanding protocols helps you choose the right tools for your workflow — whether you’re setting up a display, transferring files, building APIs, or securing user data.
Whenever you're confused about a cable, software, or service — try asking:
“What do I want this to do, and which protocol supports that?”
This cheatsheet gives you a foundation for figuring it out quickly.