LOCKEDIN / FORWARD-DEPLOYED ENGINEERING 101
Understand it.
Then build it.
You can be an experienced engineer and still be new to agents. Start with what each thing is and how it helps a person do useful work. Explore 58 original introductions across software, agents, cloud, security and delivery. Start here without a model key; each topic connects to deeper instruction and practice.
A place to start.
Choose a short reading route, or look up a concept below.
- 101UnderstandPlain language, visual walkthrough, vocabulary
- 102Build with guidanceOriginal lessons, examples and knowledge checks
- 103Practise and explainRunnable labs, failure cases and handover
Understand agents
Connect capabilities through a shared protocol while keeping permission checks in the application and server.
See the 8-concept route
For engineers who know software but are new to AI applications. Open the related definitions whenever a term is unfamiliar.
- 01The modelStart with what produces the text. Separate the model from the application using it.
- 02TokensSee the units a model reads and produces before reasoning about its input limits.
- 03ContextDecide what information reaches the model for this particular request.
- 04The agent loopPut the model inside an application that can execute permitted actions and check the result.
- 05WorkflowsSeparate a fixed sequence from a model-selected next action before choosing a framework.
- 06LangChainSee how a framework assembles the model, tools and application-controlled context.
- 07LangGraphMake the steps and shared state explicit when work needs to branch, wait or resume.
- 08MCPYou are here
Bookmark this page to return to this concept. The number shows your place in the reading route; it is not an assessment result.
An open protocol for connecting AI applications to tools and data
What is MCP?
MCP means Model Context Protocol. It defines a common way for an AI application to discover and use capabilities supplied by another program, such as reading a document or looking up a ticket. It is a communication protocol, not an AI model, a programming language or a workflow builder by itself.
An FDE needs to connect AI software to the systems where work happens. MCP can standardize that connection, while the receiving system still checks identity, permissions and the requested inputs.
UNDERSTAND THE SYSTEM
MCP connects an AI application to capabilities
Model Context Protocol (MCP) is a shared communication protocol. A host application uses an MCP client to talk to an MCP server, which exposes tools, resources or prompts.
Read the connections in plain English
- 01AI application → MCP client
The host asks its client to discover what this connected server exposes.
- 02MCP client → MCP server
Discovery can use tools/list; an allowed invocation uses tools/call with a tool name and arguments.
- 03MCP server → Procedure library
The server validates the request and the caller's access before reading the procedure.
- 04Procedure library → MCP server
The underlying service returns the permitted record or an error.
- 05MCP server → MCP client
The server sends a structured protocol response, which may also contain content.
- 06MCP client → Result in context
The client delivers the result to the host, which decides how to use it.
Check your understanding: Is MCP itself an agent or a model?
Neither. MCP standardizes communication with capabilities. The host may contain an agent and may call a model, but the protocol does not think, choose goals or grant business authority.
CONNECT IT TO WHAT YOU KNOW
Start with something familiar.
A common wire protocol lets different clients and servers understand the same kinds of messages. MCP provides that shared contract for capabilities exposed to AI applications.
MCP is a protocol. An MCP SDK is code for implementing it; an MCP server is a program exposing capabilities. None of these is itself a model or permission to perform every available action.
Words worth knowing
- Protocol
- Agreed rules for how programs exchange messages.
- Client
- The connecting component inside the AI application.
- Server
- The program that offers tools, resources or prompts through MCP.
- SDK
- A software development kit: code that helps you implement the protocol.
- Tool
- An operation with arguments, such as a procedure lookup. It may read data or change something, depending on its implementation.
- Resource
- Context the application can read using a URI, such as a document. A URI is an identifier and need not be a public web address.
- Prompt
- A reusable request template that returns messages for the application to use; retrieving it does not itself run those instructions.
Useful foundations
Open a definition and diagram here. Your lesson stays in place.
Data a program can read and checkWhat are JSON and schemas?
QUICK REFERENCE / 101
What are JSON and schemas?
JSON (JavaScript Object Notation) is a text format for exchanging data between programs. An object pairs field names with values, like a small form with named boxes. JSON Schema describes rules for those values. Neither is an AI model, and JSON works with many programming languages.
An API response, an agent's tool arguments and a model's structured answer often arrive as JSON. Before connecting them, an FDE needs to read the fields, recognize their types and explain why readable data can still be wrong for the task.
A software contractWhat is an API?
QUICK REFERENCE / 101
What is an API?
An application programming interface (API) is an agreed way for one piece of software to use another. It specifies operations, inputs, outputs and errors. Some APIs are local functions; a web API exchanges requests and responses across a network.
An agent becomes useful through connections to business systems. Those connections need clear contracts and failure behavior.
A software development kitWhat is an SDK?
QUICK REFERENCE / 101
What is an SDK?
An SDK is a collection of code and supporting tools that helps you use a platform from a programming language. A model-provider SDK can turn a Python method call into an authenticated HTTP request and parse its response. The SDK is not the model or the remote service.
FDEs integrate models, gateways and business systems. Knowing the underlying contract lets you diagnose failures when an SDK hides details.
An application built around an AI modelWhat is an AI agent?
QUICK REFERENCE / 101
What is an AI agent?
An AI agent is software that uses an AI model to choose some of the steps needed to complete a task. It can call tools, inspect their results and decide what to do next within limits set by the application. The model is one component; the agent also needs code, tools and rules about when to stop.
An FDE needs to decide when model-driven choices help a real task and when a predictable sequence is enough. An agent is useful only if it improves the work and stays within its allowed scope.
ONE EXAMPLE / FICTIONAL CASE
Follow the connections.
Instead of wiring a different ticket-lookup interface into every assistant, expose a small lookup through an MCP server. The assistant's MCP client discovers the tool and requests a call. The server decides what this caller is allowed to read.
UNDERSTAND THE SYSTEM
MCP connects an AI application to capabilities
Model Context Protocol (MCP) is a shared communication protocol. A host application uses an MCP client to talk to an MCP server, which exposes tools, resources or prompts.
See what moves and why.
Read the connections in plain English
- 01AI application → MCP client
The host asks its client to discover what this connected server exposes.
- 02MCP client → MCP server
Discovery can use tools/list; an allowed invocation uses tools/call with a tool name and arguments.
- 03MCP server → Procedure library
The server validates the request and the caller's access before reading the procedure.
- 04Procedure library → MCP server
The underlying service returns the permitted record or an error.
- 05MCP server → MCP client
The server sends a structured protocol response, which may also contain content.
- 06MCP client → Result in context
The client delivers the result to the host, which decides how to use it.
Check your understanding: Is MCP itself an agent or a model?
Neither. MCP standardizes communication with capabilities. The host may contain an agent and may call a model, but the protocol does not think, choose goals or grant business authority.
Work through it, step by step.
Discover the tool
The AI application connects through an MCP client. The MCP server describes the available lookup and its accepted inputs.
What you have now: A tool description the application can inspect
Request a lookup
The application asks to look up a ticket. The server validates the request and checks the caller's access before reading data.
What you have now: An authorized lookup or a denied request
Return the result
The server returns the permitted result. The application can use it in a response or in a larger workflow.
What you have now: A tool result; not permission for unrelated actions
See what an MCP server returns.
A maintenance assistant needs evidence and instructions before drafting a reply. Explore what the same server can offer, then choose the right capability for a task.
A model can propose the call. The application dispatches it; the server checks access.
Listing describes this capability. It has not called the lookup, read the checklist or filled the template.
Learn what is offered.
{
"name": "lookup_procedure",
"inputSchema": {
"type": "object",
"properties": {
"procedure_id": {
"type": "string",
"pattern": "^P[0-9]{1,4}$"
}
},
"required": [
"procedure_id"
],
"additionalProperties": false
}
}A descriptor is a description, not the result or permission to use it.
Authored examples · Nothing runs in this browser exercise. The linked local lab performs real SDK calls with synthetic data. Clients and servers may support only some MCP features; these are common usage patterns, not mandatory UI rules.
What does this task need?
Open the published review checklist at its known URI. Which entry in this catalog fits?
FROM RECOGNITION TO REASONING
Put the three capabilities into one useful workflow.
- 01
Describe how a person selects review instructions, the application reads the checklist and a tool looks up a permitted procedure.
- 02
Mark the separate model call that could turn those inputs into a draft. State what must be checked before using that draft.
- 03
Explain why discovering any of these capabilities does not grant access or approve sending the reply.
A connection sketch showing the template, context, lookup result and a separate model call.
Compare your reasoning when you are ready
MCP carries capabilities between programs. In this catalog, the tool returns procedure data, the resource returns checklist content and the prompt returns instructions. The host still chooses what to pass to a model and how to check its output. A server may expose the same information in different ways; choose an interface that fits its use.
Keep notes in your own working copy. Moving through these cards does not record completion.
TURN THE IDEA INTO USEFUL WORK
You can now go deeper.
MCP connects capabilities. LangGraph coordinates steps. LangChain helps assemble agents. These are different jobs and can be used together.
Academy lessons use your existing training access. Free guides stay open.
Learn with the original creators
Begin with Introducing MCP and MCP clients, then return to our local SDK lab. The provider course includes its own exercises.