memory

Documentation / agents/memory

SimpleMemory

Defined in: packages/ai-research-agent/src/agents/memory.js:57

Simple Memory Class - Core memory management functionality

Features:

• Message deduplication
• Automatic summarization
• Vector-based relevance search
• Caching with TTL
• Batch processing
• Conflict resolution

Constructors

Constructor

new SimpleMemory(
   userId: string, 
   db: any, 
   options: object): SimpleMemory;

Defined in: packages/ai-research-agent/src/agents/memory.js:70

Initialize memory system for a user

Parameters

Parameter	Type	Description
userId	string	Unique user identifier
db	any	Database connection
options	{ maxMemories: number; summaryThreshold: number; cacheExpiry: number; enableVectorSearch: boolean; enableAutoSummarization: boolean; }	Configuration options
options.maxMemories	number	Maximum memories to store
options.summaryThreshold	number	Messages before auto-summarization
options.cacheExpiry	number	Cache TTL in milliseconds
options.enableVectorSearch	boolean	Enable vector-based search
options.enableAutoSummarization	boolean	Enable auto-summarization

Returns

SimpleMemory

Properties

userId

userId: string;

Defined in: packages/ai-research-agent/src/agents/memory.js:75

db

db: any;

Defined in: packages/ai-research-agent/src/agents/memory.js:76

maxMemories

maxMemories: number;

Defined in: packages/ai-research-agent/src/agents/memory.js:77

summaryThreshold

summaryThreshold: number;

Defined in: packages/ai-research-agent/src/agents/memory.js:78

cacheExpiry

cacheExpiry: number;

Defined in: packages/ai-research-agent/src/agents/memory.js:79

batchSize

batchSize: any;

Defined in: packages/ai-research-agent/src/agents/memory.js:80

relevanceThreshold

relevanceThreshold: any;

Defined in: packages/ai-research-agent/src/agents/memory.js:81

enableVectorSearch

enableVectorSearch: boolean;

Defined in: packages/ai-research-agent/src/agents/memory.js:84

Feature flags

enableAutoSummarization

enableAutoSummarization: boolean;

Defined in: packages/ai-research-agent/src/agents/memory.js:85

recentMessages

recentMessages: any[];

Defined in: packages/ai-research-agent/src/agents/memory.js:88

State management

memoryCache

memoryCache: Map<any, any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:89

isProcessing

isProcessing: boolean;

Defined in: packages/ai-research-agent/src/agents/memory.js:90

processingQueue

processingQueue: any[];

Defined in: packages/ai-research-agent/src/agents/memory.js:91

metrics

metrics: object;

Defined in: packages/ai-research-agent/src/agents/memory.js:94

Performance metrics

Name	Type	Default value	Defined in
cacheHits	number	0	packages/ai-research-agent/src/agents/memory.js:95
cacheMisses	number	0	packages/ai-research-agent/src/agents/memory.js:96
vectorSearches	number	0	packages/ai-research-agent/src/agents/memory.js:97
summarizations	number	0	packages/ai-research-agent/src/agents/memory.js:98
errors	number	0	packages/ai-research-agent/src/agents/memory.js:99

summarizeTimeout

summarizeTimeout: Timeout;

Defined in: packages/ai-research-agent/src/agents/memory.js:162

Methods

addMessage()

addMessage(
   role: string, 
   content: string, 
   metadata: any): boolean;

Defined in: packages/ai-research-agent/src/agents/memory.js:115

Add a message to current session with intelligent deduplication

Parameters

Parameter	Type	Description
role	string	Message role ('user' or 'assistant')
content	string	Message content
metadata	any	Additional message metadata

Returns

boolean

• Whether message was added

Example

memory.addMessage('user', 'Hello, how are you?', { timestamp: Date.now() });
memory.addMessage('assistant', 'I am doing well, thank you!');

debouncedSummarize()

debouncedSummarize(): void;

Defined in: packages/ai-research-agent/src/agents/memory.js:157

Debounced summarization to prevent excessive processing

Returns

void

storeFact()

storeFact(
   content: string, 
   importance: number, 
   category: string, 
   metadata: any): Promise<string>;

Defined in: packages/ai-research-agent/src/agents/memory.js:183

Store important facts with validation and conflict resolution

Parameters

Parameter	Type	Default value	Description
content	string	undefined	Fact content
importance	number	1	Importance score (0-10)
category	string	MEMORY_TYPES.FACT	Memory category
metadata	any	{}	Additional metadata

Returns

Promise<string>

• Memory ID

Example

await memory.storeFact('User prefers dark mode', 8, 'preference', { source: 'conversation' });
await memory.storeFact('User works at Google', 9, 'personal', { confidence: 0.95 });

findSimilarFacts()

findSimilarFacts(content: string): Promise<any[]>;

Defined in: packages/ai-research-agent/src/agents/memory.js:244

Find similar facts using content similarity

Parameters

Parameter	Type	Description
content	string	Content to find similar facts for

Returns

Promise<any[]>

• Similar facts

updateMemory()

updateMemory(id: string, updates: any): Promise<void>;

Defined in: packages/ai-research-agent/src/agents/memory.js:275

Update memory record

Parameters

Parameter	Type	Description
id	string	Memory ID
updates	any	Fields to update

Returns

Promise<void>

recallRelevantMemories()

recallRelevantMemories(
   query: string, 
   limit: number, 
   options: any): Promise<any[]>;

Defined in: packages/ai-research-agent/src/agents/memory.js:303

Enhanced memory recall with caching and vector search

Parameters

Parameter	Type	Default value	Description
query	string	''	Search query
limit	number	10	Maximum results
options	any	{}	Search options

Returns

Promise<any[]>

• Relevant memories

Example

// Search for work-related memories
const workMemories = await memory.recallRelevantMemories('work', 10);

// Get most important memories
const importantMemories = await memory.recallRelevantMemories('', 5);

// Search with specific options
const recentMemories = await memory.recallRelevantMemories('meeting', 10, { 
  minImportance: 5, 
  includeMetadata: true 
});

fetchMemoriesFromDB()

fetchMemoriesFromDB(
   query: any, 
   limit: any, 
   options: any): Promise<any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:365

Fetch memories from database with filtering

Parameters

Parameter	Type
query	any
limit	any
options	any

Returns

Promise<any>

applyVectorSearch()

applyVectorSearch(
   query: any, 
   memories: any, 
   options: any): Promise<any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:395

Apply vector search to memories

Parameters

Parameter	Type
query	any
memories	any
options	any

Returns

Promise<any>

updateRelevanceScores()

updateRelevanceScores(memories: any, sentencesByRelevance: any): Promise<void>;

Defined in: packages/ai-research-agent/src/agents/memory.js:427

Update relevance scores for memories

Parameters

Parameter	Type
memories	any
sentencesByRelevance	any

Returns

Promise<void>

summarizeAndStore()

summarizeAndStore(): Promise<boolean>;

Defined in: packages/ai-research-agent/src/agents/memory.js:452

Improved summarization with error handling and batch processing

Returns

Promise<boolean>

• Success status

extractFactsFromConversation()

extractFactsFromConversation(conversationText: any): Promise<any[]>;

Defined in: packages/ai-research-agent/src/agents/memory.js:494

Extract facts from conversation using LLM

Parameters

Parameter	Type
conversationText	any

Returns

Promise<any[]>

processFactsInBatches()

processFactsInBatches(factsResponse: any): Promise<void>;

Defined in: packages/ai-research-agent/src/agents/memory.js:515

Process facts in batches to avoid overwhelming the database

Parameters

Parameter	Type
factsResponse	any

Returns

Promise<void>

clearCache()

clearCache(): void;

Defined in: packages/ai-research-agent/src/agents/memory.js:549

Clear cache utility

Returns

void

getMemoryContext()

getMemoryContext(
   query: string, 
   includeRecent: boolean, 
   options: any): Promise<string>;

Defined in: packages/ai-research-agent/src/agents/memory.js:567

Get memory context with better formatting and relevance

Parameters

Parameter	Type	Default value	Description
query	string	''	Context query
includeRecent	boolean	true	Include recent messages
options	any	{}	Context options

Returns

Promise<string>

• Formatted context

Example

const context = await memory.getMemoryContext('work meeting', true, {
  maxMemories: 5,
  minImportance: 3
});

getMetrics()

getMetrics(): object;

Defined in: packages/ai-research-agent/src/agents/memory.js:609

Get performance metrics

Returns

object

Name	Type	Default value	Defined in
cacheSize	number	‐	packages/ai-research-agent/src/agents/memory.js:612
recentMessagesCount	number	‐	packages/ai-research-agent/src/agents/memory.js:613
isProcessing	boolean	‐	packages/ai-research-agent/src/agents/memory.js:614
cacheHits	number	0	packages/ai-research-agent/src/agents/memory.js:95
cacheMisses	number	0	packages/ai-research-agent/src/agents/memory.js:96
vectorSearches	number	0	packages/ai-research-agent/src/agents/memory.js:97
summarizations	number	0	packages/ai-research-agent/src/agents/memory.js:98
errors	number	0	packages/ai-research-agent/src/agents/memory.js:99

---

MemoryAgent

Defined in: packages/ai-research-agent/src/agents/memory.js:631

Enhanced Memory Agent with better error handling and features

Features:

• Rate limiting
• Multiple LLM provider support
• Health monitoring
• Conversation management
• Memory analytics

Constructors

Constructor

new MemoryAgent(
   userId: string, 
   db: any, 
   options: object): MemoryAgent;

Defined in: packages/ai-research-agent/src/agents/memory.js:645

Initialize memory agent

Parameters

Parameter	Type	Description
userId	string	User identifier
db	any	Database connection
options	{ memoryOptions: any; defaultProvider: string; defaultApiKey: string; defaultModel: string; rateLimit: any; providers: any; }	Configuration options
options.memoryOptions	any	Memory system options
options.defaultProvider	string	Default LLM provider
options.defaultApiKey	string	Default API key
options.defaultModel	string	Default model
options.rateLimit	any	Rate limiting configuration
options.providers	any	LLM provider configurations

Returns

MemoryAgent

Properties

memory

memory: SimpleMemory;

Defined in: packages/ai-research-agent/src/agents/memory.js:650

defaultProvider

defaultProvider: string;

Defined in: packages/ai-research-agent/src/agents/memory.js:651

defaultApiKey

defaultApiKey: string;

Defined in: packages/ai-research-agent/src/agents/memory.js:652

defaultModel

defaultModel: string;

Defined in: packages/ai-research-agent/src/agents/memory.js:653

rateLimiter

rateLimiter: Map<any, any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:656

Rate limiting

rateLimitConfig

rateLimitConfig: any;

Defined in: packages/ai-research-agent/src/agents/memory.js:657

providers

providers: any;

Defined in: packages/ai-research-agent/src/agents/memory.js:663

LLM providers

sessionId

sessionId: string;

Defined in: packages/ai-research-agent/src/agents/memory.js:666

Session management

conversationHistory

conversationHistory: any[];

Defined in: packages/ai-research-agent/src/agents/memory.js:667

analytics

analytics: object;

Defined in: packages/ai-research-agent/src/agents/memory.js:670

Analytics

Name	Type	Default value	Defined in
totalMessages	number	0	packages/ai-research-agent/src/agents/memory.js:671
totalTokens	number	0	packages/ai-research-agent/src/agents/memory.js:672
averageResponseTime	number	0	packages/ai-research-agent/src/agents/memory.js:673
errorCount	number	0	packages/ai-research-agent/src/agents/memory.js:674
sessionStartTime	number	‐	packages/ai-research-agent/src/agents/memory.js:675

Methods

getDefaultProviders()

getDefaultProviders(): object;

Defined in: packages/ai-research-agent/src/agents/memory.js:682

Get default LLM providers

Returns

object

Name	Type	Defined in
groq()	(apiKey: any, model: any, temperature: any) => object	packages/ai-research-agent/src/agents/memory.js:684
openai()	(apiKey: any, model: any, temperature: any) => object	packages/ai-research-agent/src/agents/memory.js:690

generateSessionId()

generateSessionId(): string;

Defined in: packages/ai-research-agent/src/agents/memory.js:702

Generate unique session ID

Returns

string

checkRateLimit()

checkRateLimit(
   key: string, 
   maxRequests: number, 
   windowMs: number): boolean;

Defined in: packages/ai-research-agent/src/agents/memory.js:714

Rate limiting check with sliding window

Parameters

Parameter	Type	Default value	Description
key	string	undefined	Rate limit key
maxRequests	number	null	Maximum requests per window
windowMs	number	null	Window size in milliseconds

Returns

boolean

• Whether request is allowed

chat()

chat(message: string, options: any): Promise<any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:758

Main chat method with comprehensive error handling

Parameters

Parameter	Type	Description
message	string	User message
options	any	Chat options

Returns

Promise<any>

• Chat response

Example

const response = await agent.chat("Hello, I'm John", {
  provider: 'groq',
  model: 'mixtral-8x7b-32768',
  temperature: 0.7,
  systemPrompt: 'You are a helpful assistant.'
});

if (response.success) {
  console.log('Response:', response.content);
  console.log('Memory context:', response.memoryContext);
  console.log('Tokens used:', response.tokensUsed);
} else {
  console.error('Error:', response.error);
}

generateResponse()

generateResponse(prompt: any, options: any): Promise<any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:838

Generate LLM response with timeout and error handling

Parameters

Parameter	Type
prompt	any
options	any

Returns

Promise<any>

buildPrompt()

buildPrompt(
   message: any, 
   memoryContext: any, 
   options: any): string;

Defined in: packages/ai-research-agent/src/agents/memory.js:864

Build enhanced prompt with context

Parameters

Parameter	Type
message	any
memoryContext	any
options	any

Returns

string

updateAnalytics()

updateAnalytics(tokensUsed: any, responseTime: any): void;

Defined in: packages/ai-research-agent/src/agents/memory.js:894

Update analytics

Parameters

Parameter	Type
tokensUsed	any
responseTime	any

Returns

void

remember()

remember(
   fact: string, 
   importance: number, 
   category: string, 
   metadata: any): Promise<string>;

Defined in: packages/ai-research-agent/src/agents/memory.js:921

Remember a fact manually

Parameters

Parameter	Type	Default value	Description
fact	string	undefined	Fact to remember
importance	number	1	Importance score
category	string	MEMORY_TYPES.MANUAL	Memory category
metadata	any	{}	Additional metadata

Returns

Promise<string>

• Memory ID

Example

await agent.remember("User prefers meetings in the morning", 8, 'preference', {
  source: 'manual',
  confidence: 0.9
});

getMemories()

getMemories(
   query: string, 
   limit: number, 
   options: any): Promise<any[]>;

Defined in: packages/ai-research-agent/src/agents/memory.js:949

Get memories with filtering

Parameters

Parameter	Type	Default value	Description
query	string	''	Search query
limit	number	10	Maximum results
options	any	{}	Search options

Returns

Promise<any[]>

• Memories

Example

// Get all work-related memories
const workMemories = await agent.getMemories('', 20, { memoryType: 'work' });

// Search for meeting memories
const meetingMemories = await agent.getMemories('meeting', 10);

forceStoreSummary()

forceStoreSummary(): Promise<boolean>;

Defined in: packages/ai-research-agent/src/agents/memory.js:956

Force store summary of current conversation

Returns

Promise<boolean>

healthCheck()

healthCheck(): Promise<any>;

Defined in: packages/ai-research-agent/src/agents/memory.js:965

Health check for the agent

Returns

Promise<any>

• Health status

getAnalytics()

getAnalytics(): object;

Defined in: packages/ai-research-agent/src/agents/memory.js:991

Get analytics and performance metrics

Returns

object

Name	Type	Default value	Defined in
memory	{ cacheSize: number; recentMessagesCount: number; isProcessing: boolean; cacheHits: number; cacheMisses: number; vectorSearches: number; summarizations: number; errors: number; }	‐	packages/ai-research-agent/src/agents/memory.js:994
sessionId	string	‐	packages/ai-research-agent/src/agents/memory.js:995
uptime	number	‐	packages/ai-research-agent/src/agents/memory.js:996
totalMessages	number	0	packages/ai-research-agent/src/agents/memory.js:671
totalTokens	number	0	packages/ai-research-agent/src/agents/memory.js:672
averageResponseTime	number	0	packages/ai-research-agent/src/agents/memory.js:673
errorCount	number	0	packages/ai-research-agent/src/agents/memory.js:674
sessionStartTime	number	‐	packages/ai-research-agent/src/agents/memory.js:675

resetSession()

resetSession(): void;

Defined in: packages/ai-research-agent/src/agents/memory.js:1003

Reset session and clear conversation history

Returns

void

---

MEMORY_CONFIG

const MEMORY_CONFIG: object;

Defined in: packages/ai-research-agent/src/agents/memory.js:21

Configuration constants for memory management

Type declaration

Name	Type	Default value	Defined in
DEFAULT_MAX_MEMORIES	number	100	packages/ai-research-agent/src/agents/memory.js:22
DEFAULT_SUMMARY_THRESHOLD	number	10	packages/ai-research-agent/src/agents/memory.js:23
DEFAULT_CACHE_EXPIRY	number	‐	packages/ai-research-agent/src/agents/memory.js:24
DEFAULT_BATCH_SIZE	number	5	packages/ai-research-agent/src/agents/memory.js:25
DEFAULT_RELEVANCE_THRESHOLD	number	0.3	packages/ai-research-agent/src/agents/memory.js:26
DEFAULT_IMPORTANCE_RANGE	{ min: number; max: number; }	‐	packages/ai-research-agent/src/agents/memory.js:27
DEFAULT_RATE_LIMIT	{ requests: number; windowMs: number; }	‐	packages/ai-research-agent/src/agents/memory.js:28
DEFAULT_TIMEOUT	number	30000	packages/ai-research-agent/src/agents/memory.js:29
VECTOR_SEARCH_ENABLED	boolean	true	packages/ai-research-agent/src/agents/memory.js:30
AUTO_SUMMARIZATION_ENABLED	boolean	true	packages/ai-research-agent/src/agents/memory.js:31

---

MEMORY_TYPES

const MEMORY_TYPES: object;

Defined in: packages/ai-research-agent/src/agents/memory.js:37

Memory types for categorization

Type declaration

Name	Type	Default value	Defined in
FACT	string	'fact'	packages/ai-research-agent/src/agents/memory.js:38
CONVERSATION	string	'conversation'	packages/ai-research-agent/src/agents/memory.js:39
PREFERENCE	string	'preference'	packages/ai-research-agent/src/agents/memory.js:40
PERSONAL	string	'personal'	packages/ai-research-agent/src/agents/memory.js:41
WORK	string	'work'	packages/ai-research-agent/src/agents/memory.js:42
MANUAL	string	'manual'	packages/ai-research-agent/src/agents/memory.js:43

---

createMemorySchema()

function createMemorySchema(db: any): any;

Defined in: packages/ai-research-agent/src/agents/memory.js:1016

Database schema for memory system

Parameters

Parameter	Type	Description
db	any	Database instance

Returns

any

• Schema definition

---

exampleUsage()

function exampleUsage(db: any): Promise<void>;

Defined in: packages/ai-research-agent/src/agents/memory.js:1038

Comprehensive usage example with error handling

Parameters

Parameter	Type	Description
db	any	Database connection

Returns

Promise<void>