agents library

Classes

A2AAgent: An AIAgent that communicates with remote agents via the A2A protocol.
A2AAgentHandler: Bridges an AIHostAgent to the A2A (Agent2Agent) protocol.
A2AAgentOptions: Configuration options for an A2AAgent, including its identifier, name, and description.
A2AAgentSession: Session for A2A-based agents.
A2AContinuationToken: Continuation token for polling a background A2A task response.
A2AHostingJsonUtilities: Provides utility configurations for JSON serialization within the A2A hosting layer.
A2ARunDecisionContext: Provides context for a custom A2A run-mode decision.
A2AServerRegistrationOptions: Options for configuring A2A server registration.
ActionExecutor<TInput>: Handles messages by invoking an action that does not produce output.
Activity: Mirrors System.Diagnostics.Activity.
ActivityEvent: Mirrors System.Diagnostics.ActivityEvent.
ActivityNames: Activity name constants used in workflow telemetry.
ActivitySource: Mirrors System.Diagnostics.ActivitySource.
AgentAbstractionsJsonUtilities: Provides utility configurations for JSON serialization within the agent abstractions layer.
AgentClassSkill<TSelf>: Abstract base class for defining skills as Dart classes that bundle all components together.
AgentEvaluationResults: Aggregate evaluation results across multiple items.
AgentEvaluator: Batch-oriented evaluator interface for agent evaluation.
AgentFileSkill: An AgentSkill discovered from a filesystem directory backed by a SKILL.md file.
AgentFileSkillFilterContext: Provides contextual information about a discovered file to the AgentFileSkillsSourceOptions.scriptFilter and AgentFileSkillsSourceOptions.resourceFilter predicates.
AgentFileSkillResource: A file-path-backed skill resource. Reads content from a file on disk relative to the skill directory.
AgentFileSkillScript: A file-path-backed skill script.
AgentFileSkillsSource: A skill source that discovers skills from filesystem directories containing SKILL.md files.
AgentFileSkillsSourceOptions: Configuration options for file-based skill sources.
AgentFileStore: Provides an abstract base class for file storage operations.
AgentInlineSkill: A skill defined entirely in code with resources and scripts.
AgentInlineSkillContentBuilder: Internal helper that builds XML-structured content strings for code-defined and class-based skills.
AgentInlineSkillResource: A skill resource defined in code, backed by either a static value or a delegate.
AgentInlineSkillScript: A skill script backed by a delegate.
AgentInMemorySkillsSource: A skill source that holds AgentSkill instances in memory.
AgentJsonUtilities: Provides utility methods for working with JSON data in the context of agents.
AgentMcpSkill: An AgentSkill discovered from MCP skill metadata.
AgentMcpSkillResource: A skill resource backed by content returned from an MCP server.
AgentMcpSkillsSource: An AgentSkillsSource that discovers Agent Skills exposed over MCP.
AgentMcpSkillsSourceOptions: Configuration options for AgentMcpSkillsSource.
AgentMode: Represents an agent operating mode with a name and description.
AgentModeProvider: An AIContextProvider that tracks the agent's operating mode (e.g., "plan" or "execute") in the session state and provides tools for querying and switching modes.
AgentModeProviderOptions: Options controlling the behavior of AgentModeProvider.
AgentModeState: Represents the state of the agent's operating mode, stored in the session's AgentSessionStateBag.
AgentRequestMessageSourceAttribution: Attribution information for the source of an agent request message, including the component type and its identifier.
AgentRequestMessageSourceType: The source of an agent request message.
AgentResponse: Represents the response to an AIAgent run request, containing messages and metadata about the interaction.
AgentResponseEvent: Workflow output event carrying an AgentResponse.
AgentResponseOf<T>: Represents a typed response from an AIAgent run, containing both the standard AgentResponse data and a deserialized result of type T.
AgentResponseUpdate: Represents a single streaming response chunk from an AIAgent.
AgentResponseUpdateEvent: Workflow output event carrying an AgentResponseUpdate.
AgentRunContext: Provides context for an in-flight agent run.
AgentRunMode: Specifies how the A2A hosting layer decides whether to run an agent in background mode.
AgentRunOptions: Optional parameters and configuration settings for controlling agent run behavior.
AgentSession: Base abstraction for all agent threads.
AgentSessionStateBag: A key-value store for managing session-scoped state with type-safe access and JSON serialization support.
AgentSessionStateBagJsonConverter: Provides JSON encode/decode helpers for AgentSessionStateBag that delegate to AgentSessionStateBag.serialize and AgentSessionStateBag.deserialize.
AgentSessionStateBagValue: Stores a single value in an AgentSessionStateBag.
AgentSessionStateBagValueJsonConverter: Provides JSON encode/decode helpers for AgentSessionStateBagValue that delegate to AgentSessionStateBagValue.toJson and AgentSessionStateBagValue.fromJson.
AgentSessionStore: Defines the contract for storing and retrieving agent conversation threads.
AgentSkill: Abstract base class for all agent skills.
AgentSkillFrontmatter: Represents the YAML frontmatter metadata parsed from a SKILL.md file.
AgentSkillResource: Abstract base class for skill resources. A resource provides supplementary content (references, assets) to a skill.
AgentSkillScript: Abstract base class for skill scripts. A script represents an executable action associated with a skill.
AgentSkillsProvider: An AIContextProvider that exposes agent skills from one or more AgentSkillsSource instances.
AgentSkillsProviderBuilder: Fluent builder for constructing an AgentSkillsProvider backed by a composite source.
AgentSkillsProviderOptions: Configuration options for AgentSkillsProvider.
AgentSkillsSource: Abstract base class for skill sources. A skill source provides skills from a specific origin (filesystem, remote server, database, in-memory, etc.).
AgentSkillsSourceContext: Provides contextual information about the agent and session to an AgentSkillsSource when retrieving skills.
AgentWorkflowBuilder: Provides utility methods for constructing common agent workflows.
AggregateTurnMessagesExecutor: Aggregates received chat messages for release to a downstream executor.
AggregatingAgentSkillsSource: A skill source that aggregates multiple child sources, preserving their registration order.
AggregatingExecutor<T, TOutput>: Abstract base for fan-in aggregation executors.
AIAgent: Provides the base abstraction for all AI agents, defining the core interface for agent interactions and conversation management.
AIAgentBinding: Binds an AIAgent into a workflow as a chat protocol executor.
AIAgentBuilder: Provides a builder for composing a pipeline of AIAgent decorators.
AIAgentHostExecutor: Hosts an AIAgent as a workflow executor.
AIAgentHostOptions: Configuration options for hosting AI agents as workflow executors.
AIAgentIdEqualityComparer: Compares AIAgent instances by their AIAgent.id property.
AIAgentMetadata: Metadata about an AIAgent instance for identification, telemetry, and logging.
AIAgentUnservicedRequestsCollector: Scans AI response messages for unserviced tool-approval and function-call requests and submits them to registered external handlers.
AIContentExternalHandler<TRequestContent extends AIContent, TResponseContent extends AIContent>: Manages pending AI content requests and matches them to external responses.
AIContext: Additional context that AIContextProvider instances supply to enhance AI model interactions during agent invocations.
AIContextProvider: Provides an abstract base class for components that enhance AI context during agent invocations.
AIContextProviderChatClient: A delegating chat client that enriches input messages, tools, and instructions by invoking a pipeline of AIContextProvider instances before delegating to the inner chat client.
AIHostAgent: Provides a hosting wrapper around an AIAgent that adds session persistence capabilities for server-hosted scenarios.
AIJudgeLoopEvaluator: A LoopEvaluator that uses a separate judge ChatClient to decide whether the user's original request has been fully addressed, continuing the loop (with the judge's gap analysis as feedback) while the answer is "no".
AIJudgeLoopEvaluatorOptions: Configuration options for an AI-judge loop evaluator.
AlwaysApproveToolApprovalResponseContent: Wraps a ToolApprovalResponseContent with additional "always approve" settings, enabling the ToolApprovalAgent middleware to record standing approval rules so that future matching tool calls are auto-approved without user interaction.
AnonymousDelegatingAIAgent: Represents a delegating AI agent that wraps an inner agent with implementations provided by delegates.
AnthropicChatClient: A ChatClient implementation backed by Anthropic's Messages API.
AnthropicDefaults: Defaults used by the Anthropic agent integration.
AsyncRunHandle<TOutput>: A live, bidirectional handle to an in-process workflow run.
BackgroundAgentRuntimeState: Holds non-serializable runtime references for in-flight background tasks within a single parent session.
BackgroundAgentRuntimeTask: Tracks the completion state of a Dart Future while preserving the C# provider's explicit task-finalization shape.
BackgroundAgentsProvider: An AIContextProvider that enables an agent to delegate work to background agents asynchronously.
BackgroundAgentsProviderOptions: Options controlling the behavior of BackgroundAgentsProvider.
BackgroundAgentState: Represents the serializable state of background tasks managed by the BackgroundAgentsProvider, stored in the session's AgentSessionStateBag.
BackgroundTaskCompletionLoopEvaluator: A LoopEvaluator that keeps re-invoking the wrapped agent until a BackgroundAgentsProvider reports that no background tasks are still running.
BackgroundTaskCompletionLoopEvaluatorOptions: Provides configuration options for BackgroundTaskCompletionLoopEvaluator.
BackgroundTaskInfo: Represents the metadata and result of a background task managed by the BackgroundAgentsProvider.
BooleanProgressLedgerSlot: A boolean-valued ProgressLedgerSlot.
CachingAgentSkillsSource: A skill source decorator that caches the result of the inner source's getSkills call, returning the cached list on subsequent invocations.
CachingAgentSkillsSourceOptions: Options for configuring a CachingAgentSkillsSource.
CallResult: Captures the result of invoking an executor.
ChatClientAgent: Provides an AIAgent that delegates to a ChatClient implementation.
ChatClientAgentContinuationToken: Represents a continuation token for ChatClientAgent operations that carries additional context needed to resume interrupted responses.
ChatClientAgentOptions: Configuration options for a ChatClientAgent.
ChatClientAgentRunOptions: Provides specialized run options for ChatClientAgent instances, extending the base agent run options with chat-specific configuration.
ChatClientAgentSession: Provides a conversation session for use with ChatClientAgent.
ChatForwardingExecutor: Forwards chat protocol messages to another workflow executor.
ChatHistoryMemoryProvider: A context provider that stores all chat history in a vector store and is able to retrieve related chat history later to augment the conversation.
ChatHistoryMemoryProviderOptions: Options controlling the behavior of ChatHistoryMemoryProvider.
ChatHistoryMemoryProviderScope: Allows scoping of chat history for the ChatHistoryMemoryProvider.
ChatHistoryProvider: Provides an abstract base class for fetching and storing chat history messages for use during agent execution.
ChatMessageJsonConverter: Provides JSON encode/decode helpers for framework ChatMessages.
ChatProtocol: Describes the chat message protocol used by agent workflow executors.
ChatProtocolExecutor: Workflow executor that invokes an AIAgent using the chat protocol.
ChatProtocolExecutorOptions: Configuration for ChatProtocolExecutor.
ChatReducerCompactionStrategy: A compaction strategy that delegates to a ChatReducer to reduce the conversation's included messages.
Checkpoint: Represents a durable workflow checkpoint.
CheckpointableRun: Base contract for run handles that support checkpointing.
CheckpointInfo: Represents a checkpoint with a unique identifier.
CheckpointInfoConverter: Provides serialization and dictionary-key support for CheckpointInfo values.
CheckpointingHandle: Provides checkpointing operations for a running workflow.
CheckpointManager: Minimal checkpoint manager contract for workflow run APIs.
CheckpointManagerImpl: Default checkpoint manager backed by a CheckpointStore.
CheckpointStore: Stores durable workflow checkpoints.
CollectingWorkflowContext: Minimal in-memory workflow context used by tests and lightweight executors.
CompactionMessageGroup: Represents a logical group of ChatMessage instances that must be kept or removed together during compaction.
CompactionMessageIndex: A collection of CompactionMessageGroup instances and derived metrics based on a flat list of ChatMessage objects.
CompactionProvider: A AIContextProvider that applies a CompactionStrategy to compact the message list before each agent invocation.
CompactionStrategy: Base class for strategies that compact a CompactionMessageIndex.
CompactionStrategyChatReducer: A ChatReducer adapter that delegates to a CompactionStrategy.
CompactionTelemetry: Shared telemetry names for compaction operations.
CompactionTriggers: Factory methods for CompactionTrigger predicates.
CompletionMarkerLoopEvaluator: A LoopEvaluator that stops the loop once a configured marker string appears in the agent's latest response, and otherwise continues with feedback asking the agent to keep working and to emit the marker when done.
CompletionMarkerLoopEvaluatorOptions: Configuration options for a completion-marker loop evaluator.
ConcurrentEndExecutor: Aggregates output messages from concurrent agents.
ConcurrentEventSink: An EventSink that forwards events synchronously to a callback.
ConcurrentWorkflowBuilder: Fluent builder for concurrent agent workflows: a fan-out start that broadcasts the incoming messages to every participating agent, a per-agent accumulator that batches each agent's outgoing messages, and a fan-in aggregator that reduces them into a single output list.
Configured<T>: Pairs a workflow object with configuration.
ConfiguredExecutorBinding: Executor binding decorated with configuration metadata.
ContainerUser: UID/GID pair passed to docker run --user.
ContextWindowCompactionStrategy: A compaction strategy that derives token thresholds from a model's context window size and maximum output tokens, applying tool-result eviction before truncation.
ConversationSplitter: Strategy for splitting a conversation into query and response halves for evaluation.
ConversationSplitters: Built-in conversation splitters for common evaluation patterns.
DeduplicatingAgentSkillsSource: A skill source decorator that removes duplicate skills by name, keeping only the first occurrence.
DelayedDeserialization: Supports lazy deserialization of a value when the target type is not known until retrieval time.
DelegateLoopEvaluator: A LoopEvaluator that delegates the re-invocation decision and feedback to a user-supplied callback.
DelegatingAgentSessionStore: Provides an abstract base class for agent session stores that delegate operations to an inner store instance while allowing for extensibility and customization.
DelegatingAgentSkillsSource: Provides an abstract base class for skill sources that delegate operations to an inner source while allowing for extensibility and customization.
DelegatingAIAgent: Abstract base class for AI agents that delegate operations to an inner agent while allowing extensibility via the decorator pattern.
DeliveryMapping: Maps one or more MessageEnvelopes to one or more executor targets.
DirectEdgeData: Edge data for a direct source-to-target connection.
DirectEdgeInfo: Serializable direct edge information.
DirectEdgeRunner: Routes a message across a direct edge.
DockerNetworkMode: Well-known values for the network parameter on DockerShellExecutor.
DockerShellExecutor: A ShellExecutor that runs commands inside a Docker container.
DockerShellExecutorOptions: Configuration options for DockerShellExecutor.
Edge: Represents a workflow edge.
EdgeConnection: Connects an edge to executable routing behavior.
EdgeData: Base data for a workflow edge.
EdgeId: Identifies a workflow edge.
EdgeIdConverter: Provides serialization and dictionary-key support for EdgeId values.
EdgeInfo: Serializable workflow edge information.
EdgeMap: Indexes workflow edges by their source executor identifiers.
EdgeRunner: Base class for all workflow edge runners.
EnvironmentSanitizer: Helpers for the cleanEnvironment mode where the spawned shell does not inherit the parent process environment — except for a small allowlist of variables that the shell needs to locate itself and basic tools.
EvalCheckResult: Result of a single check on a single evaluation item.
EvalChecks: Built-in check functions for common evaluation patterns.
EvalItem: Provider-agnostic data for a single evaluation item.
EvalItemResult: Per-item result from a Foundry evaluation run, with individual evaluator scores and error details.
EvalScoreResult: A single evaluator's score on one evaluation item.
EventNames: Event name constants used in workflow telemetry.
EventSink: Receives workflow events and dispatches them to registered listeners.
Executor<TInput, TOutput>: Base type for workflow executors.
ExecutorBinding: Binds an executor identity to a runtime executor instance factory.
ExecutorCompletedEvent: Event emitted when an executor completes.
ExecutorConfig: Configuration applied while creating or binding an executor.
ExecutorEvent: Base type for events associated with a workflow executor.
ExecutorFailedEvent: Event emitted when an executor fails.
ExecutorIdentity: Identifies an executor instance by an optional case-insensitive string id.
ExecutorIdentityConverter: Provides serialization and dictionary-key support for ExecutorIdentity values.
ExecutorInfo: Serializable executor registration information.
ExecutorInstanceBinding: Binds a workflow executor to an existing shared instance.
ExecutorInvokedEvent: Event emitted when an executor is invoked.
ExecutorOptions: Options that describe runtime behavior for an executor registration.
ExecutorPlaceholder: Placeholder binding used while a workflow is being assembled.
ExpectedToolCall: A tool call that an agent is expected to make.
ExternalRequest<TRequest, TResponse>: Represents a request that must be satisfied outside the workflow runtime.
ExternalRequestContext<TRequest, TResponse>: Tracks an outstanding external request and its eventual response.
ExternalRequestSink: Accepts external requests from executors and delivers their responses.
ExternalResponse<TResponse>: Represents a response to an ExternalRequest.
FanInEdgeData: Edge data for multiple source executors synchronized into one target.
FanInEdgeInfo: Serializable fan-in edge information.
FanInEdgeRunner: Waits for every fan-in source before routing an aggregate message.
FanInEdgeState: Accumulates messages from multiple fan-in sources, releasing all of them as a grouped batch once every source has contributed.
FanOutEdgeData: Edge data for a source executor connected to multiple targets.
FanOutEdgeInfo: Serializable fan-out edge information.
FanOutEdgeRunner: Routes one source message to every fan-out target.
FileAccessProvider: An AIContextProvider that provides file access tools to an agent for writing, reading, deleting, listing, searching, and editing files.
FileAccessProviderOptions: Options controlling the behavior of FileAccessProvider.
FileEditor: Helpers shared by the file access and file memory providers for the replace and replace_lines tools.
FileLineEdit: Represents a single whole-line replacement used by the file access and file memory replace_lines tools.
FileListEntry: Represents a file entry returned by the FileMemoryProvider list files tool, containing the file name and an optional description.
FileMemoryProvider: An AIContextProvider that provides file-based memory tools to an agent for storing, retrieving, modifying, listing, deleting, and searching files.
FileMemoryProviderOptions: Options controlling the behavior of FileMemoryProvider.
FileMemoryState: Represents the state of the FileMemoryProvider, stored in the session's AgentSessionStateBag.
FileSearchMatch: Represents a match found within a file during a search operation.
FileSearchResult: Represents a result from searching files, containing the file name, a content snippet, and matching lines.
FileStoreEntry: Represents a single direct child of a directory in an AgentFileStore, returned by AgentFileStore.listChildrenAsync.
FileSystemAgentFileStore: A file-system-backed implementation of AgentFileStore that stores files on disk under a configurable root directory.
FileSystemJsonCheckpointStore: File-system JSON checkpoint store.
FilteringAgentSkillsSource: A skill source decorator that filters skills using a caller-supplied predicate.
ForwardedOptions: State passed from this instance into the inner agent, circumventing the intermediate OpenTelemetryChatClient.
ForwardingChatClient: The stub ChatClient used to delegate from OpenTelemetryChatClient into the inner AIAgent.
FullSplitter: Splits a conversation at the first user message, returning everything from the start through that message as the query half.
FunctionEvaluator: Factory for creating EvalCheck delegates from typed lambda functions.
FunctionExecutor<TInput, TOutput>: Handles messages by invoking a user-provided Dart callback.
FunctionInvocationDelegatingAgent: Internal agent decorator that adds function invocation middleware logic.
FunctionStatefulExecutor<TState, TInput, TOutput>: Stateful executor backed by a single callback.
GeminiChatClient: A ChatClient implementation backed by the Gemini API.
GeminiClient: Minimal REST client for the Gemini API.
GeminiDefaults: Defaults used by the Gemini agent integration.
GeneratedEvaluatorRef: A reference to a generated rubric evaluator that already exists in the provider's registry.
GroupChatHost: Hosts a group chat workflow and selects participating agents with a manager.
GroupChatManager: A manager that manages the flow of a group chat.
GroupChatWorkflowBuilder: Provides a builder for specifying group chat relationships between agents.
HandlerRegistry: A registry that collects typed message handlers during ReflectingExecutor.configureHandlers.
HandoffAgentExecutor: Utilities for agent executors used by handoff workflows.
HandoffAgentExecutorOptions: Options used by handoff agent executors.
HandoffEndExecutor: Terminates a handoff workflow by emitting the accumulated conversation history from conversation.
HandoffMessagesFilter: Filters handoff function call messages from history before target handoff.
HandoffStartExecutor: Executor used at the start of a handoff workflow.
HandoffState: Represents routing state in a handoff workflow.
HandoffTarget: Describes a handoff to a specific target AIAgent.
HandoffWorkflowBuilder: Provides a builder for specifying handoff relationships between agents.
HarnessAgent: A pre-configured agent that wraps a ChatClient with function invocation, per-service-call chat history persistence, compaction, and the default harness context providers.
HarnessAgentOptions: Configuration options for a HarnessAgent.
HeadTailBuffer: Bounded accumulator that keeps the first half of the input and the most recent half (rolling tail), summing to cap UTF-8 bytes total.
HostedAgentBuilder: Represents a builder for configuring AI agents within a hosting environment.
HostedWorkflowBuilder: Represents a builder for configuring workflows within a hosting environment.
Identified: Identifies a workflow component by stable string identifier.
InMemoryAgentFileStore: An in-memory implementation of AgentFileStore that stores files in a dictionary.
InMemoryAgentSessionStore: In-memory implementation of AgentSessionStore for development and testing scenarios.
InMemoryChatHistoryProvider: In-memory implementation of ChatHistoryProvider with optional message reduction support.
InMemoryChatHistoryProviderOptions: Configuration options for InMemoryChatHistoryProvider.
InMemoryChatHistoryProviderState: The persisted state for InMemoryChatHistoryProvider.
InMemoryCheckpointManager: In-memory checkpoint manager.
InProcessExecutionEnvironment: Executes workflows in the current Dart isolate.
InProcessExecutionOptions: Configuration options for in-process workflow execution.
InProcessRunner: In-process implementation of SuperStepRunnable.
InProcessRunnerContext: Orchestrates in-process workflow execution: instantiates executors, routes messages through edges, manages state, and participates in superstep joins.
InProcExecutionEnvironment: In-process execution environment that uses InProcessRunner.
InProcStepTracer: StepTracer implementation for in-process workflow execution.
InputWaiter: Synchronises the workflow run loop with external input events.
InvokedContext: Context passed to AIContextProvider.invoked.
InvokingContext: Context passed to ChatHistoryProvider.invoking and AIContextProvider.invoking at the start of an agent invocation.
IsolationKeyScopedAgentSessionStore: A delegating agent session store that scopes session keys by an isolation key provided by a SessionIsolationKeyProvider, ensuring that sessions are isolated per logical partition (e.g., user, tenant, or composite key).
IsolationKeyScopedAgentSessionStoreOptions: Options for configuring an IsolationKeyScopedAgentSessionStore.
JsonCheckpointStore: In-memory JSON checkpoint store.
JsonConverter<T>: Mirrors C# System.Text.Json.Serialization.JsonConverter<T>.
JsonConverterBase<T>: Abstract base for JSON converters that serialize and deserialize a specific type T.
JsonConverterDictionarySupportBase<T>: Extends JsonConverterBase with support for using T values as JSON dictionary keys via stringify / parse, and provides escape / unescape helpers that double | characters to avoid collision with the |-based composite key format.
JsonElement: Mirrors C# System.Text.Json.JsonElement (a parsed JSON value). In Dart, JSON values are plain Dart objects from dart:convert.
JsonMarshaller: Serializes JSON-shaped checkpoint values deterministically.
JsonSerializer: Mirrors C# System.Text.Json.JsonSerializer.
JsonSerializerOptions: Stub types for C# System.Text.Json types. These are placeholders until a proper Dart serialization strategy is implemented. The intent is preserved from the original C# source. Mirrors C# System.Text.Json.JsonSerializerOptions.
JsonTypeInfo<T>: Mirrors C# System.Text.Json.Serialization.Metadata.JsonTypeInfo<T>.
JsonWireSerializedValue: Stores a JSON-shaped value together with optional type metadata, supporting lazy deserialization when the target type is determined later.
LastTurnSplitter: Splits a conversation at the last user message, returning everything from the start through that message as the query half.
LocalEvaluator: Evaluator that runs check functions locally without API calls.
LocalShellExecutor: A ShellExecutor that runs commands directly on the host machine.
LocalShellExecutorOptions: Configuration options for LocalShellExecutor.
LockstepRunEventStream: A RunEventStream that drives one superstep at a time.
LoggingAgent: A delegating AIAgent that logs agent operations to a Logger.
LoopAgent: A DelegatingAIAgent that re-invokes the wrapped agent in a loop until the configured LoopEvaluator set decides to stop.
LoopAgentOptions: Configuration options for a loop agent.
LoopContext: The per-run state that a LoopEvaluator uses to decide whether a loop agent should re-invoke the wrapped agent and what feedback to provide.
LoopEvaluation: The result produced by a LoopEvaluator after an agent iteration: whether the loop should re-invoke the wrapped agent and, optionally, the feedback or explicit messages that should inform the next iteration.
LoopEvaluator: The abstract base class for the component that decides, after each agent iteration, whether a loop agent should re-invoke the wrapped agent and what feedback to provide.
MagenticOrchestrator: Centralized, re-entrant Magentic orchestrator.
MagenticOrchestratorEvent: Base type for Magentic orchestration events.
MagenticPlanCreatedEvent: Signals creation of the initial plan.
MagenticPlanReviewRequest: Request for human review of a proposed Magentic plan.
MagenticPlanReviewResponse: Review feedback for a proposed Magentic plan.
MagenticProgressLedger: Maintains a ledger of progress made by a Magentic workflow.
MagenticProgressLedgerUpdatedEvent: Signals an update to the progress ledger during a coordination round.
MagenticReplannedEvent: Signals creation of a new plan in response to a stall.
MagenticWorkflowBuilder: Fluent builder for Magentic One multi-agent orchestration workflows.
McpClientAIFunction: An AIFunction backed by an MCP tool.
McpSkillIndex: DTO for skill://index.json.
McpSkillIndexEntry: A single skill discovery index entry.
McpTaskOptions: Options that control task-augmented MCP tool invocation.
MeaiEvaluatorAdapter: Adapter that wraps an MEAI Evaluator into an AgentEvaluator.
MessageAIContextProvider: Abstract base class for components that enhance AI context during agent invocations by supplying additional chat messages.
MessageAIContextProviderAgent: A delegating AI agent that enriches input messages by invoking a pipeline of MessageAIContextProvider instances before delegating to the inner agent, and notifies those providers after the inner agent completes.
MessageDelivery: Represents a queued message delivery.
MessageEnvelope: Wraps a workflow message with routing metadata.
MessageHandlerInfo: Associates a messageType with a type-erased handler callback.
MessageInjectingChatClient: A delegating chat client that supports injecting messages into the function execution loop.
MessageInvokingContext: Context passed to MessageAIContextProvider.invoking.
MessageMerger: Merges streams of AgentResponseUpdates grouped by response and message identifiers into coherent AgentResponse objects.
MessageRouter: Routes executor outputs across workflow edges.
MiddlewareEnabledFunction: Wraps an AIFunction to inject middleware logic on invocation.
MultiPartyConversation: Stores the shared conversation history for a multi-party workflow.
NonApprovalRequiredFunctionBypassingChatClient: A delegating chat client that automatically removes ToolApprovalRequestContent for tools that do not actually require approval, storing auto-approved results in the session for transparent re-injection on the next request.
NonThrowingChannelReaderAsyncEnumerable<T>: Wraps a Stream and absorbs OperationCanceledExceptions so iteration completes normally when cancelled rather than propagating the exception.
NoopAgentSessionStore: An AgentSessionStore that does not persist sessions.
NullRedactor: A Redactor that returns the original value unchanged.
OpenTelemetryAgent: Provides a delegating AIAgent implementation that instruments agent operations via OpenTelemetryChatClient.
OpenTelemetryAgentTags: OpenTelemetry attribute name constants for the agent identity dimension.
OpenTelemetryConsts: Provides constants used by the agent telemetry services.
OpenTelemetryGenAI: OpenTelemetry attribute name constants for generative-AI operations.
OpenTelemetryOperation: OpenTelemetry attribute name constants for the operation dimension.
OpenTelemetryProvider: OpenTelemetry attribute name constants for the AI provider dimension.
OrchestrationBuilderBase<TBuilder extends OrchestrationBuilderBase<TBuilder>>: Common fluent surface shared by every orchestration-style workflow builder: human-readable name + description, and the withOutputFrom / withIntermediateOutputFrom output-designation pair with memoized defaults-suppression semantics.
OutputFilter: Decides whether an executor is authorised to emit workflow output.
OutputMessagesExecutor: Batches received chat messages and publishes them as the final result.
OutputTag: Identifies the kind of output that a WorkflowOutputEvent represents.
OutputTagJsonConverter: Provides JSON encode/decode helpers for OutputTag that map a tag to and from its string OutputTag.value.
PerEvaluatorResult: Per-evaluator pass/fail breakdown from an evaluation run.
PerServiceCallChatHistoryPersistingChatClient: A delegating chat client that persists chat history and updates session state after each individual service call within the function-invoking loop.
PipelineCompactionStrategy: A compaction strategy that executes a sequential pipeline of CompactionStrategy instances against the same CompactionMessageIndex.
PortableMessageEnvelope: Serializable form of a workflow message envelope.
PortableValue: Wraps an arbitrary value alongside its TypeId, supporting optional lazy deserialization via DelayedDeserialization.
PortableValueConverter: Provides JSON serialization for PortableValue with support for lazy deserialization of the inner value.
PortBinding: Base class for bindings associated with named ports.
PortHandlerExecutor<TRequest, TResponse>: A workflow executor that responds to external requests arriving on port.
ProgressLedgerSlot: A single question slot in a MagenticProgressLedger.
ProtocolBuilder: Builder used by executors to describe their accepted and produced messages.
ProtocolDescriptor: Describes the message protocol supported by an executor or workflow.
ProviderSessionState<TState>: Provides strongly-typed state management for providers, enabling reading and writing of provider-specific state to/from an AgentSession's AgentSessionStateBag.
Redactor: Redacts sensitive values before they are written to diagnostic logs.
ReflectingExecutor: Base executor class that dispatches inbound messages to typed handlers.
ReplacingRedactor: A Redactor that replaces every value with a fixed replacement string.
RequestHaltEvent: Event emitted when a workflow requests that the run halt.
RequestInfoEvent: Event triggered when a workflow executor requests external information.
RequestInfoExecutor<TRequest extends Object, TResponse extends Object>: Executor that routes typed requests to an ExternalRequestSink and returns the external response.
RequestPort<TRequest, TResponse>: Describes an external request/response port available to an executor.
RequestPortBinding<TRequest, TResponse>: Binds a request port to a Dart callback.
RequestPortDescriptor: Non-generic description for a RequestPort.
RequestPortInfo: Serializable request port description.
ResettableExecutor: Implemented by executors that can be reset between workflow runs.
ResolvedShell: A resolved shell binary together with its argv-building methods.
ResponseEdgeRunner: Routes an external response envelope back to the executor that issued the corresponding request.
RoundRobinGroupChatManager: Provides a GroupChatManager that selects agents in round-robin order.
RouteBuilder: Fluent route builder rooted at a workflow executor.
RubricScore: A single dimension's score from a rubric-based evaluator run.
Run: Represents a workflow run that tracks status and emitted events.
RunEventStream: Abstract base for WorkflowEvent event streams.
RunnerContext: Shared context for an in-process workflow runner.
RunnerStateData: Mutable state carried by an in-process workflow runner.
ScopeId: A unique identifier for a scope within an executor.
ScopeKey: A unique key within a specific scope, combining a ScopeId and a key string.
ScopeKeyConverter: Provides serialization and dictionary-key support for ScopeKey values.
SequentialWorkflowBuilder: Fluent builder for sequential agent workflows: a pipeline where the output of one agent is the input to the next, terminating in an aggregator that yields the accumulated ChatMessages as the workflow output.
SessionCheckpointCache: Tracks checkpoints by workflow session.
SessionIsolationKeyProvider: Provides an abstract base class for resolving session isolation keys used to scope agent sessions.
ShellEnvironmentProvider: An AIContextProvider that probes the shell environment and injects instructions describing the OS, shell version, and available tools.
ShellEnvironmentProviderOptions: Configuration options for ShellEnvironmentProvider.
ShellEnvironmentSnapshot: A point-in-time snapshot of the shell environment the agent is using.
ShellExecutor: Pluggable backend that runs shell commands on behalf of a tool.
ShellOutputChunk: A live line of output produced while a shell command is running.
ShellPolicy: Evaluates shell commands against configurable allow/deny regex patterns.
ShellPolicyOutcome: The result of a ShellPolicy evaluation.
ShellRequest: A shell command awaiting a policy decision.
ShellResolver: Resolves which shell binary to use and provides argv builders for both stateless and persistent invocation modes.
ShellResult: The outcome of a single shell command invocation.
ShellSession: Manages a persistent shell subprocess using a sentinel protocol to bracket each command's stdout and capture its exit code.
SlidingWindowCompactionStrategy: A compaction strategy that removes the oldest user turns and their associated response groups to bound conversation length.
StatefulExecutor<TState, TInput, TOutput>: Base class for executors that maintain state across message handling.
StatefulExecutorOptions<TState>: Options for stateful executors.
StateManager: Manages workflow execution state using a two-phase write pattern.
StateScope: Stores key/value state for a single scope, keyed by ScopeId.
StateUpdate: An immutable record of a single state change — either an upsert or a deletion.
StepContext: Workflow context for a single executor invocation.
StepTracer: Emits traces for key step lifecycle events within a workflow execution.
StorePathGlobMatcher: Minimal case-insensitive glob matcher for file names.
StorePaths: Internal helper for normalizing and validating relative store paths and matching glob patterns. Shared across AgentFileStore implementations.
StreamingAggregators: Provides streaming aggregation functions for processing input sequences in a stateful, incremental manner.
StreamingRun: A workflow run instance supporting streaming workflow events.
StreamingRunEventStream: A RunEventStream backed by a broadcast StreamController.
StringProgressLedgerSlot: A string-valued ProgressLedgerSlot.
SubworkflowBinding: Binds a sub-Workflow as a WorkflowHostExecutor inside a parent workflow.
SubworkflowErrorEvent: Event emitted when a subworkflow encounters an error.
SubworkflowWarningEvent: Event emitted when a subworkflow encounters a warning condition.
SummarizationCompactionStrategy: A compaction strategy that uses an LLM to summarize older portions of the conversation.
SuperStepCompletedEvent: Event triggered when a SuperStep completes.
SuperStepCompletionInfo: Debug information about the SuperStep that finished running.
SuperStepEvent: Base class for SuperStep-scoped events.
SuperStepJoinContext: Context that a sub-workflow runner uses to communicate with its parent.
SuperStepRunnable: Minimal interface for a runner that participates in a superstep.
SuperStepRunner: Executes queued workflow messages in SuperStep order.
SuperStepStartedEvent: Event triggered when a SuperStep starts.
SuperStepStartInfo: Debug information about the SuperStep starting to run.
SwitchBuilder: Fluent builder for routing one source executor to multiple targets.
Tags: Telemetry tag key constants for workflow observability.
TaskAwareMcpClientAIFunction: An MCP tool function that must be invoked through task augmentation.
TextSearchProvider: A text search context provider that performs a search over external knowledge and injects formatted results into the AI invocation context, or exposes a search tool for on-demand use.
TextSearchProviderOptions: Options controlling the behavior of TextSearchProvider.
TextSearchProviderState: Represents the per-session state of a TextSearchProvider.
TextSearchResult: Represents a single retrieved text search result.
TodoCompleteInput: Input for completing a single todo item via TodoProvider.
TodoCompletionLoopEvaluator: A LoopEvaluator that keeps re-invoking the wrapped agent until a TodoProvider has no remaining (incomplete) todo items, optionally only while the agent is operating in one of a configured set of modes tracked by an AgentModeProvider.
TodoCompletionLoopEvaluatorOptions: Configuration options for a todo-completion loop evaluator.
TodoItem: Represents a single todo item managed by the TodoProvider.
TodoItemInput: Represents the input for creating a new todo item via the TodoProvider.
TodoProvider: An AIContextProvider that provides todo management tools and instructions to an agent for tracking work items during long-running complex tasks.
TodoProviderOptions: Options controlling the behavior of TodoProvider.
TodoState: Represents the state of the todo list managed by the TodoProvider, stored in the session's AgentSessionStateBag.
Tokenizer: Provides token counting for compaction metrics.
ToolApprovalAgent: Middleware that handles standing tool-approval rules and queues approval requests so callers see at most one unresolved request at a time.
ToolApprovalAgentOptions: Options for configuring the ToolApprovalAgent middleware.
ToolApprovalRule: Represents a standing approval rule for automatically approving tool calls without requiring explicit user approval each time.
ToolApprovalState: Represents the persisted state of standing tool approval rules, stored in the session's AgentSessionStateBag.
ToolResultCompactionStrategy: A compaction strategy that collapses old tool call groups into single concise assistant messages.
TruncationCompactionStrategy: A compaction strategy that removes the oldest non-system message groups, keeping at least minimumPreservedGroups most-recent groups intact.
TurnToken: Sent to an AIAgent-based executor to request a response to accumulated chat messages.
TypeId: Identifies a serialized Dart type in checkpoint wire data.
UpdateKey: Identifies an update within a specific scope.
Utf8JsonReader: Mirrors C# System.Text.Json.Utf8JsonReader.
Utf8JsonWriter: Mirrors C# System.Text.Json.Utf8JsonWriter.
ValueTaskTypeErasure: Converts a typed MessageHandlerCallback<T> to a type-erased handler.
WireMarshaller: Converts values to and from checkpoint wire values.
Workflow: A class that represents a workflow that can be executed.
WorkflowBuilder: Builder for creating a Workflow from executor bindings and edges.
WorkflowCatalog: Provides a catalog of registered workflows within the hosting environment.
WorkflowChatHistoryProvider: A ChatHistoryProvider that persists chat messages in the agent session state bag and supports bookmark-based incremental retrieval.
WorkflowContext: Context supplied to an executor while handling a workflow message.
WorkflowErrorEvent: Event emitted for a workflow error.
WorkflowEvent: Base type for workflow events.
WorkflowExecutionEnvironment: Defines an execution environment for running and streaming workflows.
WorkflowHostAgent: Hosts a Workflow behind the AIAgent abstraction.
WorkflowHostAgentSession: Agent session used by WorkflowHostAgent.
WorkflowHostExecutor: Hosts a sub-Workflow as an executor inside a parent workflow.
WorkflowInfo: Serializable workflow definition information.
WorkflowInfoOutputExecutorsConverter: JSON converter for WorkflowInfo.outputExecutors that supports both the new map shape ({ "id": ["intermediate"] }) and the legacy array shape (["id1", "id2"]). Legacy-shaped payloads are read as if every id had been registered as a regular (untagged) output source; output is always written in the new map shape.
WorkflowOutputEvent: Event emitted when a workflow yields output.
WorkflowSession: Represents a workflow session.
WorkflowsJsonUtilities: JSON serialization helpers for workflow checkpoints.
WorkflowStartedEvent: Event emitted when a workflow starts.
WorkflowStateContext: Workflow context extended with state read/write support and sub-workflow attachment for in-process execution.
WorkflowTelemetryContext: No-op telemetry context.
WorkflowTelemetryOptions: Configuration options for workflow telemetry.
WorkflowVisualizer: Generates visual representations of a Workflow.
WorkflowWarningEvent: Event emitted for a non-fatal workflow warning.

Enums

ActivityKind: Mirrors System.Diagnostics.ActivityKind.
ActivityStatusCode: Mirrors System.Diagnostics.ActivityStatusCode.
BackgroundTaskStatus: Represents the status of a background task managed by the BackgroundAgentsProvider.
ChatReducerTriggerEvent: Defines when a ChatReducer is triggered in InMemoryChatHistoryProvider.
CompactionGroupKind: Identifies the kind of a CompactionMessageGroup.
EdgeRunnerDeliveryStatus: Describes the outcome of a message delivery attempt through an edge runner.
ExecutionMode: Controls how workflow events are streamed during execution.
HandoffToolCallFilteringBehavior: Specifies how tool calls are filtered from handoff workflow history.
RunStatus: Specifies the current operational state of a workflow run.
SearchBehavior: Behavior choices for the provider.
ShellFamily: Identifies the shell family the agent is talking to.
ShellMode: Specifies how a shell executor dispatches commands to the underlying shell.
ShellOutputChannel: Identifies the process stream that produced a ShellOutputChunk.
TextSearchBehavior: Behavior choices for the provider.
ToolCalledMode: Specifies how a tool-name list matches called tools.

Extensions

A2AAgentCardExtensions on A2AAgentCard: Extension methods for A2AAgentCard that create AIAgent instances.
A2AClientExtensions on A2AClient: Extension methods for A2AClient that create A2AAgent instances.
A2AMetadataToAdditionalPropertiesExtension on A2ASV: Converts A2A metadata to an agent additional-properties dictionary.
A2AServerHostedAgentBuilderExtensions on HostedAgentBuilder: Registers an A2A server bridge for an agent configured through a HostedAgentBuilder.
A2AServerServiceCollectionExtensions on ServiceCollection: Registers A2A server bridges (an A2AAgentExecutor per agent) in a ServiceCollection.
AdditionalPropertiesExtensions on AdditionalPropertiesDictionary: Extension methods for AdditionalPropertiesDictionary that use the runtime type name of T as the dictionary key.
AdditionalPropertiesToA2AMetadataExtension on AdditionalPropertiesDictionary: Converts an agent additional-properties dictionary to A2A metadata.
AgentEvaluationExtensions on AIAgent: Extension methods for evaluating agents and responses.
AgentHostingServiceCollectionExtensions on ServiceCollection: Provides extension methods for configuring AI agents in a service collection.
AgentResponseExtensions on AgentResponse: Extension methods for AgentResponse and AgentResponseUpdate.
AgentResponseToPartsExtension on AgentResponse: Converts an AgentResponse to A2A A2APart objects.
AgentResponseUpdateExtensions on AgentResponseUpdate: Extension methods for AgentResponseUpdate.
AgentResponseUpdateIterableExtensions on Iterable<AgentResponseUpdate>: Extension methods for iterables of AgentResponseUpdate.
AgentResponseUpdateStreamExtensions on Stream<AgentResponseUpdate>: Extension methods for streams of AgentResponseUpdate.
AgentResponseUpdateToPartsExtension on AgentResponseUpdate: Converts an AgentResponseUpdate to A2A A2APart objects.
AgentSessionExtensions on AgentSession: Extension methods for AgentSession.
AgentSkillsProviderBuilderMcpExtensions on AgentSkillsProviderBuilder: MCP-specific extensions for AgentSkillsProviderBuilder.
AIAgentExtensions on AIAgent: Provides extensions for AIAgent.
AIAgentStructuredOutputExtensions on AIAgent: Extension methods on AIAgent for structured (typed) output.
AIContextProviderChatClientBuilderExtensions on ChatClientBuilder: Provides extension methods for adding AIContextProvider support to ChatClientBuilder instances.
AIWorkflowAbstractionsExtensions on Workflow: Extensions for exposing workflows through the AI agent abstractions.
AnthropicClientAgentExtensions on AnthropicClient: Extension methods for creating chat clients and agents from Anthropic clients.
AsyncRunHandleExtensions on AsyncRunHandle<TOutput>: Convenience extensions on AsyncRunHandle.
ChatClientAgentRunOptionsExtension on ChatClientAgent: Convenience extension methods on ChatClientAgent for typed run options.
ChatClientBuilderAgentExtensions on ChatClientBuilder: Provides extension methods for building ChatClientAgent instances from ChatClientBuilder pipelines.
ChatClientExtensions on ChatClient: Provides extension methods for creating ChatClientAgent instances from ChatClient pipelines.
ChatClientHarnessExtensions on ChatClient: Provides extension methods for creating a HarnessAgent from a ChatClient.
ChatMessageContentEquality on ChatMessage?: Content-based equality comparison for ChatMessage instances.
ChatMessageExtensions on ChatMessage: Extension methods for ChatMessage related to agent request message source attribution.
ChatMessageListTextExtensions on List<ChatMessage>: Concatenates ChatMessage.text values across a list of messages.
ChatMessagesToPartsExtension on Iterable<ChatMessage>: Converts a collection of ChatMessage objects to A2A A2APart objects.
ChatStrategyExtensions on CompactionStrategy: Provides extension methods for CompactionStrategy.
CheckpointInfoKey on CheckpointInfo: Converts CheckpointInfo instances to string keys for use in maps.
ConfigurationExtensions on Configured<TSubject>: Extension methods for upcasting Configured values.
ConfiguredExtensions on T: Configuration helpers.
EdgeRepresentation on Edge: Converts Edge instances to their checkpoint representation.
EdgeRunnerDeliveryStatusExtension on EdgeRunnerDeliveryStatus: String-value helpers for EdgeRunnerDeliveryStatus.
ExecutorBindingExtensions on ExecutorBinding: Convenience methods for executor bindings.
ExecutorBindingRepresentation on ExecutorBinding: Converts ExecutorBinding instances to their checkpoint representation.
ExecutorInstanceBindingExtensions on Executor: Convenience methods for executor instances.
ExternalRequestExtensions on ExternalRequest<TRequest, TResponse>: Extension methods for ExternalRequest.
FunctionInvocationDelegatingAgentBuilderExtensions on AIAgentBuilder: Provides extension methods for configuring and customizing AIAgentBuilder instances.
GeminiClientAgentExtensions on GeminiClient: Extension methods for creating chat clients and agents from Gemini clients.
HostApplicationBuilderAgentExtensions on HostApplicationBuilder: Provides extension methods for configuring AI agents in a host application builder.
HostApplicationBuilderWorkflowExtensions on HostApplicationBuilder: Provides extension methods for configuring AI workflows in a host application builder.
HostedAgentBuilderExtensions on HostedAgentBuilder: Provides extension methods for configuring HostedAgentBuilder.
HostedWorkflowBuilderExtensions on HostedWorkflowBuilder: Provides extension methods for HostedWorkflowBuilder to enable additional workflow configuration scenarios.
LoggingAgentBuilderExtensions on AIAgentBuilder: Provides extension methods for adding logging support to AIAgentBuilder instances.
MapTryExtensions on Map<K, V>: Extension methods to port .NET ConcurrentDictionary / Dictionary patterns to Dart's Map API.
McpClientTaskExtensions on McpClient: Extensions for exposing MCP tools as extensions AI functions.
OpenAIAgentResponseExtensions on AgentResponse: OpenAI conversion helpers for AgentResponse.
OpenAIAgentResponseUpdateStreamExtensions on Stream<AgentResponseUpdate>: OpenAI conversion helpers for Streams of AgentResponseUpdate.
OpenAIChatClientAgentExtensions on OpenAIChatClient: Convenience methods for creating ChatClientAgent instances backed by OpenAIChatClient.
OpenTelemetryAgentBuilderExtensions on AIAgentBuilder: Provides extension methods for adding OpenTelemetry instrumentation to AIAgentBuilder instances.
OpenTelemetryWorkflowExecutionEnvironmentExtensions on WorkflowExecutionEnvironment: Adds OpenTelemetry instrumentation to a WorkflowExecutionEnvironment.
ReflectionExtensions on WorkflowBuilder: WorkflowBuilder extensions for ReflectingExecutor.
RequestPortDescriptorRepresentation on RequestPortDescriptor: Converts RequestPortDescriptor instances to their checkpoint representation.
RequestPortExtensions on RequestPort<TRequest, TResponse>: Extension helpers for matching ExternalResponse to a RequestPort.
RouteBuilderExtensions on WorkflowBuilder: WorkflowBuilder edge-routing extensions for ReflectingExecutor.
ToolApprovalAgentBuilderExtensions on AIAgentBuilder: Provides extension methods for adding tool approval middleware to AIAgentBuilder instances.
ToolApprovalRequestContentExtensions on ToolApprovalRequestContent?: Provides extension methods on ToolApprovalRequestContent for creating AlwaysApproveToolApprovalResponseContent instances that instruct the ToolApprovalAgent middleware to record standing approval rules.
WorkflowBuilderExtensions on WorkflowBuilder: C#-style workflow builder convenience methods.
WorkflowContextExtensions on WorkflowContext: Extension methods for working with WorkflowContext instances.
WorkflowEvaluationExtensions on WorkflowExecutionEnvironment: Evaluation helpers for WorkflowExecutionEnvironment.
WorkflowHostingExtensions on Workflow: Extension methods for hosting a Workflow as an AIAgent.
WorkflowOutputEventExtensions on WorkflowOutputEvent: Extension helpers for inspecting WorkflowOutputEvent tag membership.
WorkflowRepresentation on Workflow: Converts Workflow instances to their checkpoint representation.
WorkflowSpanExtensions on Span: Extension methods on Span for workflow-specific attribute and status helpers.
WorkflowTracerExtensions on Tracer: Extension methods on Tracer for creating workflow-specific spans.

Constants

inProcessExecution → const InProcessExecutionEnvironment: Default in-process workflow execution environment.
inProcExecution → const InProcExecutionEnvironment: Default InProcExecutionEnvironment instance.
localHistoryConversationId → const String: A sentinel conversation ID used to signal to FunctionInvokingChatClient that the conversation is service-managed (for internal bookkeeping only — never sent to the underlying service).

Functions

activityEvent(String name, {Map<String, Object?> tags = const {}}) → ActivityEvent: Factory function to create an ActivityEvent (mirrors C# constructor call pattern new ActivityEvent(name, ...)).
checkpointStringList(Object? value) → List<String>: Converts a JSON list into a string list.
replacingRedactor(String replacement) → Redactor

Typedefs

Action = void Function(): A zero-argument void callback.
Action1<T> = void Function(T): A single-argument void callback.
Action2<T1, T2> = void Function(T1, T2): A two-argument void callback.
Action3<T1, T2, T3> = void Function(T1, T2, T3): A three-argument void callback.
Action4<T1, T2, T3, T4> = void Function(T1, T2, T3, T4): A four-argument void callback.
AgentFileSkillScriptRunner = Future<Object?> Function(AgentFileSkill skill, AgentFileSkillScript script, JsonElement? arguments, ServiceProvider? serviceProvider, {CancellationToken? cancellationToken}): Function for running file-based skill scripts.
CatchAllFn = Future<CallResult?> Function(Object message, WorkflowContext context, CancellationToken cancellationToken): A catch-all handler for messages whose exact type has no registered handler.
ChatHistoryMemoryProviderState = State: Public alias for State, which the package barrel hides because its name collides with Flutter's widget State.
CompactionTrigger = bool Function(CompactionMessageIndex index): Defines a condition based on CompactionMessageIndex metrics.
ConfigureA2AServerOptions = void Function(A2AServerRegistrationOptions options): A callback that configures A2AServerRegistrationOptions.
EvalCheck = EvalCheckResult Function(EvalItem item): Function for a synchronous evaluation check on a single item.
Func<TIn, TOut> = TOut Function(TIn): Generic function type aliases ported from C# delegate conventions.
Func2<T1, T2, TOut> = TOut Function(T1, T2): A function that takes two arguments and returns TOut.
Func3<T1, T2, T3, TOut> = TOut Function(T1, T2, T3): A function that takes three arguments and returns TOut.
Func4<T1, T2, T3, T4, TOut> = TOut Function(T1, T2, T3, T4): A function that takes four arguments and returns TOut.
FunctionInvocationDelegateFunc = Func4<AIAgent, FunctionInvocationContext, Func2<FunctionInvocationContext, CancellationToken, Future<Object?>>, CancellationToken, Future<Object?>>: Delegate type used to intercept and customise AI function invocations.
LoopEvaluateCallback = Future<LoopEvaluation> Function(LoopContext context, {CancellationToken? cancellationToken}): A callback that decides whether to re-invoke the agent and what feedback to provide, given the full LoopContext.
MessageHandlerCallback<T> = Future<Object?> Function(T message, WorkflowContext context, CancellationToken? cancellationToken): A callback that handles a single message of type T.
MessageHandlerFn = Future<CallResult?> Function(Object message, WorkflowContext context, CancellationToken cancellationToken): A function that handles a typed message and returns a CallResult.
RunAgentDelegate = Future<AgentResponse> Function(Iterable<ChatMessage> messages, AgentSession? session, AgentRunOptions? options, AIAgent innerAgent, CancellationToken? cancellationToken): Function type for a custom non-streaming agent run delegate.
RunInBackgroundCallback = Future<bool> Function(A2ARunDecisionContext context, CancellationToken? cancellationToken): A delegate that decides whether an agent run should be wrapped in a background A2A task.
RunStreamingAgentDelegate = Stream<AgentResponseUpdate> Function(Iterable<ChatMessage> messages, AgentSession? session, AgentRunOptions? options, AIAgent innerAgent, CancellationToken? cancellationToken): Function type for a custom streaming agent run delegate.
SessionCreatedCallback = Future<void> Function(AgentSession session, CancellationToken? cancellationToken): A callback invoked whenever the loop agent creates a new session, so the caller can capture the latest session.
SharedAgentRunDelegate = Future<AgentResponse> Function(Iterable<ChatMessage> messages, AgentSession? session, AgentRunOptions? options, Future<AgentResponse> innerInvoker(Iterable<ChatMessage>, AgentSession?, AgentRunOptions?, CancellationToken?), CancellationToken? cancellationToken): Function type for a shared delegate that handles both run and streaming.
ShellOutputCallback = void Function(ShellOutputChunk chunk): Receives live shell output as a command executes.
TextSearchAsync = Future<Iterable<TextSearchResult>> Function(String input, CancellationToken cancellationToken): Async delegate that performs a text search and returns an iterable of TextSearchResult instances.
ToolCallFormatter = String Function(CompactionMessageGroup group): Formats a CompactionMessageGroup representing a tool-call exchange into a concise summary string.

Exceptions / Errors

ShellCommandRejectedException: Thrown by a ShellExecutor when a ShellPolicy denies a command.

agents library

Classes

Enums

Extensions

Constants

Functions

Typedefs

Exceptions / Errors

agents package

agents library