Class AIAgent<TInput, TOutput>`Abstract`

Base class for AI Agents in the Stratix framework.

An AI Agent is an aggregate root that encapsulates AI capabilities, manages its own state, memory, and execution context.

Example

class CustomerSupportAgent extends AIAgent<SupportTicket, SupportResponse> {
  readonly name = 'Customer Support Agent';
  readonly description = 'Handles customer support tickets';
  readonly capabilities = [AgentCapabilities.CUSTOMER_SUPPORT, 'ticket_routing'];
  readonly model = {
    provider: 'anthropic',
    model: 'claude-3-sonnet',
    temperature: 0.7,
    maxTokens: 2000
  };

  async execute(ticket: SupportTicket): Promise<AgentResult<SupportResponse>> {
    // Implementation
  }
}

Type Parameters

TInput
The type of input the agent accepts
TOutput
The type of output the agent produces

Hierarchy (View Summary)

AggregateRoot<"AIAgent">
- AIAgent
  - StreamableAgent

Accessors

id

get id(): EntityId<T>
Gets the unique identifier of this entity.

Returns EntityId<T>
Inherited from AggregateRoot.id
- Defined in core/src/core/Entity.ts:41

createdAt

get createdAt(): Date
Gets the timestamp when this entity was created.

Returns Date
Inherited from AggregateRoot.createdAt
- Defined in core/src/core/Entity.ts:48

updatedAt

get updatedAt(): Date
Gets the timestamp when this entity was last updated.

Returns Date
Inherited from AggregateRoot.updatedAt
- Defined in core/src/core/Entity.ts:55

Constructors

`Protected`constructor

new AIAgent<TInput, TOutput>(
    _id: EntityId<"AIAgent">,
    _createdAt: Date,
    _updatedAt: Date,
): AIAgent<TInput, TOutput>
Type Parameters
- TInput
- TOutput
Parameters
- _id: EntityId<"AIAgent">
- _createdAt: Date
- _updatedAt: Date
Returns AIAgent<TInput, TOutput>
Inherited from AggregateRoot.constructor
- Defined in core/src/core/Entity.ts:32

Methods

`Protected` `Abstract`execute

execute(input: TInput): Promise<AgentResult<TOutput>>
Protected
Main execution method for the agent. Must be implemented by concrete agent classes.

This method is called internally by executeWithEvents() and should contain the core agent logic. Domain events are automatically recorded by the wrapper.
Parameters
- input: TInput
  The input data for the agent
Returns Promise<AgentResult<TOutput>>
A promise resolving to the agent result
- Defined in core/src/ai-agents/AIAgent.ts:105

executeWithEvents

executeWithEvents(input: TInput): Promise<AgentResult<TOutput>>
Executes the agent with automatic domain event recording. This is the public method that should be called by orchestrators and clients.

Records the following events:
- AgentExecutionStarted: When execution begins
- AgentExecutionCompleted: When execution succeeds
- AgentExecutionFailed: When execution fails
Parameters
- input: TInput
  The input data for the agent
Returns Promise<AgentResult<TOutput>>
A promise resolving to the agent result
- Defined in core/src/ai-agents/AIAgent.ts:119

`Protected` `Optional`beforeExecute

beforeExecute(input: TInput): Promise<void>
Optional hook called before execution
Parameters
- input: TInput
  The input that will be executed
Returns Promise<void>
- Defined in core/src/ai-agents/AIAgent.ts:184

`Protected` `Optional`afterExecute

afterExecute(result: AgentResult<TOutput>): Promise<void>
Optional hook called after successful execution
Parameters
- result: AgentResult<TOutput>
  The execution result
Returns Promise<void>
- Defined in core/src/ai-agents/AIAgent.ts:191

`Protected` `Optional`onError

onError(error: Error): Promise<void>
Optional hook called when an error occurs
Parameters
- error: Error
  The error that occurred
Returns Promise<void>
- Defined in core/src/ai-agents/AIAgent.ts:198

setContext

setContext(context: AgentContext): void
Sets the execution context for this agent
Parameters
- context: AgentContext
  The execution context
Returns void
- Defined in core/src/ai-agents/AIAgent.ts:205

setMemory

setMemory(memory: AgentMemory): void
Sets the memory system for this agent
Parameters
- memory: AgentMemory
  The memory implementation
Returns void
- Defined in core/src/ai-agents/AIAgent.ts:215

setRetryConfig

setRetryConfig(config: RetryConfig): void
Sets the retry configuration for this agent
Parameters
- config: RetryConfig
  The retry configuration
Returns void
- Defined in core/src/ai-agents/AIAgent.ts:224

setTimeout

setTimeout(ms: number): void
Sets the execution timeout for this agent
Parameters
- ms: number
  Timeout in milliseconds
Returns void
- Defined in core/src/ai-agents/AIAgent.ts:233

executeWithRetry

executeWithRetry(input: TInput): Promise<AgentResult<TOutput>>
Executes the agent with automatic retry logic
Parameters
- input: TInput
  The input data for the agent
Returns Promise<AgentResult<TOutput>>
A promise resolving to the agent result
- Defined in core/src/ai-agents/AIAgent.ts:246

`Protected`remember

remember(key: string, value: unknown, type?: "short" | "long"): Promise<void>
Stores a value in agent memory
Parameters
- key: string
  The key to store under
- value: unknown
  The value to store
- type: "short" | "long" = 'short'
  Memory type: 'short' for session, 'long' for persistent
Returns Promise<void>
- Defined in core/src/ai-agents/AIAgent.ts:307

`Protected`recall

recall(key: string): Promise<unknown>
Retrieves a value from agent memory
Parameters
- key: string
  The key to retrieve
Returns Promise<unknown>
The stored value, or null if not found
- Defined in core/src/ai-agents/AIAgent.ts:325

`Protected`searchMemory

searchMemory(query: string, limit?: number): Promise<unknown[]>
Searches memory semantically
Parameters
- query: string
  The search query
- limit: number = 5
  Maximum results to return
Returns Promise<unknown[]>
Array of relevant values
- Defined in core/src/ai-agents/AIAgent.ts:339

`Protected`forget

forget(type?: "short" | "long" | "all"): Promise<void>
Removes a value from memory
Parameters
- type: "short" | "long" | "all" = 'short'
  Type of memory to clear
Returns Promise<void>
- Defined in core/src/ai-agents/AIAgent.ts:351

getAgentId

getAgentId(): AgentId
Gets the agent's unique identifier

Returns AgentId
- Defined in core/src/ai-agents/AIAgent.ts:361

hasCapability

hasCapability(capability: string): boolean
Checks if this agent has a specific capability
Parameters
- capability: string
  The capability to check for
Returns boolean
- Defined in core/src/ai-agents/AIAgent.ts:370

toMetadata

toMetadata(): {
    id: string;
    name: string;
    description: string;
    version: string;
    capabilities: string[];
    model: ModelConfig;
}
Gets agent metadata as a plain object

Returns {
    id: string;
    name: string;
    description: string;
    version: string;
    capabilities: string[];
    model: ModelConfig;
}
- Defined in core/src/ai-agents/AIAgent.ts:377

`Protected`recordToolUsage

recordToolUsage(
    toolName: string,
    toolArguments: Record<string, unknown>,
    toolResult?: unknown,
): void
Protected
Records a tool used event Should be called by subclasses when they use tools
Parameters
- toolName: string
  Name of the tool used
- toolArguments: Record<string, unknown>
  Arguments passed to the tool
- OptionaltoolResult: unknown
  Result from the tool execution
Returns void
- Defined in core/src/ai-agents/AIAgent.ts:464

`Protected`record

record(event: DomainEvent): void
Protected
Records a domain event that occurred within this aggregate. Events are stored internally and can be retrieved later for publishing.
Parameters
- event: DomainEvent
  The domain event to record
Returns void
Example
```
this.record({
  occurredAt: new Date(),
  userId: this.id.value,
  email: 'user@example.com'
});
```
Inherited from AggregateRoot.record
- Defined in core/src/core/AggregateRoot.ts:58

pullDomainEvents

pullDomainEvents(): DomainEvent[]
Retrieves and clears all domain events recorded by this aggregate. This is typically called after the aggregate is persisted, to publish the events.

Returns DomainEvent[]
An array of domain events
Example
```
await repository.save(order);
const events = order.pullDomainEvents();
await eventBus.publish(events);
```
Inherited from AggregateRoot.pullDomainEvents
- Defined in core/src/core/AggregateRoot.ts:75

hasDomainEvents

hasDomainEvents(): boolean
Checks if this aggregate has any recorded domain events.

Returns boolean
true if there are pending domain events, false otherwise
Inherited from AggregateRoot.hasDomainEvents
- Defined in core/src/core/AggregateRoot.ts:86

`Protected`touch

touch(): void
Protected
Updates the updatedAt timestamp to the current time. Should be called whenever the entity's state changes.

Returns void
Inherited from AggregateRoot.touch
- Defined in core/src/core/Entity.ts:65

equals

equals(other: Entity<"AIAgent">): boolean
Compares this entity with another for equality based on identity. Two entities are equal if they have the same ID and are of the same type.
Parameters
- other: Entity<"AIAgent">
  The other entity to compare with
Returns boolean
true if the entities are equal, false otherwise
Inherited from AggregateRoot.equals
- Defined in core/src/core/Entity.ts:76

Properties

`Abstract` `Readonly`name

name: string

Human-readable name of the agent

`Abstract` `Readonly`description

description: string

Description of what the agent does

`Abstract` `Readonly`version

version: AgentVersion

`Abstract` `Readonly`capabilities

capabilities: string[]

Capabilities this agent has

`Abstract` `Readonly`model

model: ModelConfig

LLM model configuration

`Protected` `Optional`_context

_context?: AgentContext

Current execution context

`Protected` `Optional`_memory

_memory?: AgentMemory

`Protected` `Optional`retryConfig

retryConfig?: RetryConfig

`Protected` `Optional`timeoutMs

timeoutMs?: number

Class AIAgent<TInput, TOutput>Abstract

Example

Type Parameters

Hierarchy (View Summary)

Index

Accessors

Constructors

Methods

Properties

Accessors

id

Returns EntityId<T>

createdAt

Returns Date

updatedAt

Returns Date

Constructors

Protectedconstructor

Type Parameters

Parameters

Returns AIAgent<TInput, TOutput>

Methods

Protected Abstractexecute

Parameters

Returns Promise<AgentResult<TOutput>>

executeWithEvents

Parameters

Returns Promise<AgentResult<TOutput>>

Protected OptionalbeforeExecute

Parameters

Returns Promise<void>

Protected OptionalafterExecute

Parameters

Returns Promise<void>

Protected OptionalonError

Parameters

Returns Promise<void>

setContext

Parameters

Returns void

setMemory

Parameters

Returns void

setRetryConfig

Parameters

Returns void

setTimeout

Parameters

Returns void

executeWithRetry

Parameters

Returns Promise<AgentResult<TOutput>>

Protectedremember

Parameters

Returns Promise<void>

Protectedrecall

Parameters

Returns Promise<unknown>

ProtectedsearchMemory

Parameters

Returns Promise<unknown[]>

Protectedforget

Parameters

Returns Promise<void>

getAgentId

Returns AgentId

hasCapability

Parameters

Returns boolean

toMetadata

Returns { id: string; name: string; description: string; version: string; capabilities: string[]; model: ModelConfig;}

ProtectedrecordToolUsage

Parameters

Returns void

Protectedrecord

Parameters

Returns void

Example

pullDomainEvents

Returns DomainEvent[]

Class AIAgent<TInput, TOutput>`Abstract`

`Protected`constructor

`Protected` `Abstract`execute

`Protected` `Optional`beforeExecute

`Protected` `Optional`afterExecute

`Protected` `Optional`onError

`Protected`remember

`Protected`recall

`Protected`searchMemory

`Protected`forget

Returns {
id: string;
name: string;
description: string;
version: string;
capabilities: string[];
model: ModelConfig;
}

`Protected`recordToolUsage

`Protected`record

`Protected`touch

`Abstract` `Readonly`name

`Abstract` `Readonly`description

`Abstract` `Readonly`version

`Abstract` `Readonly`capabilities

`Abstract` `Readonly`model

`Protected` `Optional`_context

`Protected` `Optional`_memory

`Protected` `Optional`retryConfig

`Protected` `Optional`timeoutMs