Key details for this exam, checked against the published exam outline
Each question shows the correct answer and an explanation of why it is right
A customer service automation team is comparing two systems: System A follows a fixed decision tree that always asks the same sequence of questions and cannot deviate from predefined branches, while System B uses an LLM that can dynamically decide which tools to call, in what order, and can revise its plan based on intermediate results. What best characterizes the key distinction between System A and System B?
AI agents are distinguished from rule-based workflows and traditional chatbots by autonomy, reasoning, and dynamic tool use. System B's ability to plan, decide on tool invocation order, and adapt based on intermediate results is the hallmark of agentic behavior. System A's fixed decision tree reflects a rule-based workflow, not an agent, regardless of its use of conditional branches. The other options misattribute agentic characteristics.
A developer is building an agent that must look up current inventory levels via an API, reason about whether the requested quantity is available, and then decide whether to place an order or notify the user of a shortage, iterating through several intermediate reasoning-and-action steps before producing a final answer. Which reasoning pattern best matches this behavior?
ReAct (Reasoning and Acting) explicitly interleaves reasoning traces with actions (tool calls) and observations, allowing the agent to adapt its plan based on real-time feedback like API results. Chain-of-Thought alone describes step-by-step reasoning without necessarily incorporating external tool actions and observation loops. Prompt chaining and zero-shot prompting do not capture the iterative reasoning-action-observation loop described.
A team is building an agent using LangChain that needs to search a knowledge base, call a weather API, and format responses consistently. They define a ChatModel, a set of Tool objects wrapping the API and search functions, and a PromptTemplate to instruct the model on how to reason about which tool to invoke. Which statement correctly describes the role of these LangChain abstractions in the agent construction?
In LangChain, the ChatModel provides the LLM-based reasoning, Tools wrap external functions or APIs that the agent can call, and Prompts provide structured instructions and context that guide the model's decisions, including which tool to select. These abstractions work together within an agent executor loop to enable dynamic tool-augmented reasoning; the other options misstate the actual roles of these components.
An enterprise architecture team wants to connect a single AI agent to multiple external systems, such as a CRM database and a document repository, using a standardized protocol so tool integrations can be added without custom point-to-point code for each system. They plan to expose these systems as MCP servers accessible over Streamable HTTP, communicating using JSON-RPC 2.0 messages. Is this an appropriate use of the Model Context Protocol (MCP) for this scenario?
MCP is designed exactly for this purpose: it standardizes how AI agents (as MCP clients/hosts) communicate with external tools, resources, and prompts exposed by MCP servers, using JSON-RPC 2.0 as the message format over transports such as stdio or Streamable HTTP. This lets the agent integrate with the CRM and document repository through a consistent protocol rather than custom integrations for each system, making the statement true.
A development team is building a customer support solution with the OpenAI Agents SDK. They want a triage agent to receive incoming requests, and based on the topic, transfer control of the conversation to either a billing specialist agent or a technical support specialist agent, each with its own tools and instructions. They also want to ensure that any agent output claiming a refund amount is validated before being returned to the user. Which combination of Agents SDK primitives should they use to implement this design?
In the OpenAI Agents SDK, Handoffs are the primitive specifically designed to transfer control of a conversation from one agent to another specialized agent based on context, which fits the triage-to-specialist routing requirement. Guardrails are used to validate inputs, outputs, or agent actions, such as checking that refund amounts are valid before returning them to the user. Combining Handoffs for routing and Guardrails for output validation matches the intended multi-agent design pattern; the other options either misuse Tools for routing, ignore validation, or misunderstand what Guardrails and Runner do.
49 questions covering all exam domains, starting from $20
Exam domains verified against: Official Oracle 1Z0-1157-26 exam guide, last checked September 2026.
Distinguish AI agents from traditional chatbots and rule-based systems based on autonomy, reasoning capability, and tool use. Understand how LLM, Tools, and Orchestration loop work together as core agent components, and learn reasoning patterns like Chain-of-Thought and ReAct for guiding agent behavior. Study safety considerations and guardrail techniques to control agent actions and prevent unwanted outputs.
Learn the core LangChain abstractions of chat models, prompts, tools, and agents, and how each contributes to building agents. Practice constructing AI agents by applying LangChain tools, prompts, and chains in realistic scenarios. Trace the reasoning and tool execution flow within a LangChain agent to understand how decisions are made and actions are taken.
Sample question from this domain above: Q3
Understand how MCP standardizes and simplifies the integration between AI agents and external tools and services. Study MCP core components including hosts, clients, servers, tools, resources, and prompts, and their roles in agent-tool integration. Learn the JSON-RPC 2.0 message format and the stdio and Streamable HTTP transport options to implement MCP in agentic workflows.
Sample question from this domain above: Q4
Learn how the OpenAI Responses API supports agentic applications and extends agent capabilities. Study the Agents SDK primitives of Agent, Runner, Tool, Handoffs, and Guardrails and how they work together in agentic workflows. Apply function calling and tools to build capable agents, and understand how multi-agent design patterns and handoffs route work between specialized agents. Examine how guardrails validate inputs, outputs, and agent actions to control behavior.
Sample question from this domain above: Q5
Identify OCI services that support the full lifecycle of enterprise AI agents from development to execution and scaling. Learn how the OCI Enterprise AI Agents service enables agent development, orchestration, and runtime execution. Study the building blocks including the Responses API, tools, memory, and vector stores, and practice building and running a basic agent. Explore deployment and scaling options for production agentic workloads.
Learn how Oracle AI Database supports agentic AI workloads through Oracle AI Vector Search, Select AI, and MCP integration. Understand vector concepts including VECTOR data type, vector embeddings, and similarity search, and trace the workflow from document chunking through embedding generation to retrieval. Apply Vector Search to ground agent responses with relevant enterprise data. Study how the Private Agent Factory enables no-code agent creation and how Select AI enables natural language interaction with database data.
Common questions about the exam itself
