Agent Configuration Reference¶

All fields available on Agent.builder().

Field	Type	Required	Default	Description
`role`	`String`	Yes	--	The agent's role/title. Used in prompts and logging.
`goal`	`String`	Yes	--	The agent's primary objective. Included in the system prompt.
`background`	`String`	No	`null`	Persona context for the system prompt. Omitted when null or blank.
`tools`	`List<Object>`	No	`[]`	Tools available to this agent. Each entry must implement `AgentTool` or have `@Tool`-annotated methods.
`llm`	`ChatModel`	Yes	--	Any LangChain4j `ChatModel`.
`allowDelegation`	`boolean`	No	`false`	When `true`, a `delegate` tool is auto-injected at execution time.
`verbose`	`boolean`	No	`false`	When `true`, prompts and LLM responses are logged at INFO level.
`maxIterations`	`int`	No	`25`	Maximum number of tool-call iterations before the agent is forced to produce a final answer. Must be greater than zero.
`responseFormat`	`String`	No	`""`	Extra formatting instructions appended to the system prompt.

Validation¶

The following validations are applied at build() time:

role must not be null or blank
goal must not be null or blank
llm must not be null
maxIterations must be greater than zero
Each tool must implement AgentTool or have at least one @Tool-annotated method

Example¶

Agent agent = Agent.builder()
    .role("Senior Research Analyst")
    .goal("Find accurate, well-structured information on any topic")
    .background("You are a veteran researcher with 15 years of technology industry experience.")
    .tools(List.of(new WebSearchTool()))
    .llm(model)
    .allowDelegation(true)
    .verbose(false)
    .maxIterations(25)
    .responseFormat("Always structure your response with clear headings.")
    .build();

Immutability¶

Agent is an immutable value object. Use toBuilder() to create modified copies:

Agent verboseAgent = agent.toBuilder().verbose(true).build();