所有方法 实例方法 具体方法 已过时的方法
| 限定符和类型 |
方法和说明 |
ReActAgent |
build() |
ReActAgent.Builder |
defaultInterceptorAdd(int index,
ReActInterceptor val) |
ReActAgent.Builder |
defaultInterceptorAdd(ReActInterceptor... vals) |
ReActAgent.Builder |
defaultSkillAdd(Skill... skills) |
ReActAgent.Builder |
defaultSkillAdd(Skill skill,
int index) |
ReActAgent.Builder |
defaultToolAdd(FunctionTool... tools) |
ReActAgent.Builder |
defaultToolAdd(java.lang.Iterable<FunctionTool> tools) |
ReActAgent.Builder |
defaultToolAdd(java.lang.Object toolObj) |
ReActAgent.Builder |
defaultToolAdd(ToolProvider toolProvider) |
ReActAgent.Builder |
defaultToolContextPut(java.util.Map<java.lang.String,java.lang.Object> objectsMap) |
ReActAgent.Builder |
defaultToolContextPut(java.lang.String key,
java.lang.Object value) |
ReActAgent.Builder |
description(java.lang.String val)
|
ReActAgent.Builder |
enablePlanning(boolean val)
|
ReActAgent.Builder |
feedbackDescription(java.util.function.Function<ReActTrace,java.lang.String> provider) |
ReActAgent.Builder |
feedbackDescription(java.lang.String description) |
ReActAgent.Builder |
feedbackMode(boolean val)
反馈模式（允许主动寻求外部帮助/反馈）
|
ReActAgent.Builder |
feedbackReasonDescription(java.util.function.Function<ReActTrace,java.lang.String> provider) |
ReActAgent.Builder |
feedbackReasonDescription(java.lang.String description) |
ReActAgent.Builder |
finishMarker(java.lang.String val)
定义 LLM 输出中的任务结束标识符
|
ReActAgent.Builder |
graphAdjuster(java.util.function.Consumer<org.noear.solon.flow.GraphSpec> graphBuilder)
微调推理图结构
|
ReActAgent.Builder |
instruction(java.util.function.Function<ReActTrace,java.lang.String> instruction) |
ReActAgent.Builder |
instruction(java.lang.String instruction) |
ReActAgent.Builder |
maxSteps(int val)
单次任务允许的最大推理步数（防止死循环）
|
ReActAgent.Builder |
maxStepsExtensible(boolean val) |
ReActAgent.Builder |
maxStepsLimit(int val) |
ReActAgent.Builder |
modelOptions(java.util.function.Consumer<ModelOptionsAmend<?,ReActInterceptor>> chatOptions) |
ReActAgent.Builder |
name(java.lang.String name) |
ReActAgent.Builder |
outputKey(java.lang.String val) |
ReActAgent.Builder |
outputSchema(java.lang.String val) |
ReActAgent.Builder |
outputSchema(java.lang.reflect.Type type) |
ReActAgent.Builder |
planInstruction(java.util.function.Function<ReActTrace,java.lang.String> provider)
|
ReActAgent.Builder |
planningInstruction(java.util.function.Function<ReActTrace,java.lang.String> provider) |
ReActAgent.Builder |
planningInstruction(java.lang.String instruction) |
ReActAgent.Builder |
planningMode(boolean val)
规划模式（推理前先制定计划）
|
ReActAgent.Builder |
profile(AgentProfile profile) |
ReActAgent.Builder |
profile(java.util.function.Consumer<AgentProfile> profileConsumer) |
ReActAgent.Builder |
retryConfig(int maxRetries,
long retryDelayMs) |
ReActAgent.Builder |
role(java.lang.String role) |
ReActAgent.Builder |
sessionWindowSize(int val) |
ReActAgent.Builder |
style(ReActStyle style) |
ReActAgent.Builder |
systemPrompt(java.util.function.Consumer<ReActSystemPrompt.Builder> promptBuilder)
已过时。
|
ReActAgent.Builder |
systemPrompt(ReActSystemPrompt val)
已过时。
|
ReActAgent.Builder |
then(java.util.function.Consumer<ReActAgent.Builder> consumer) |