agentdb-memory-patterns

Name: agentdb-memory-patterns
Author: spencermarx

301stars24forks

How to install agentdb-memory-patterns?

spencermarx/open-code-review/agentdb-memory-patterns

$npx -y skills add spencermarx/open-code-review --skill agentdb-memory-patterns

Installs into the current project.

Summary

Implement persistent memory patterns for AI agents using AgentDB. Includes session memory, long-term storage, pattern learning, and context management. Use when building stateful agents, chat systems, or intelligent assistants.

Files · 1

View on GitHub

SKILL.md

1	# AgentDB Memory Patterns
2
3	## What This Skill Does
4
5	Provides memory management patterns for AI agents using AgentDB's persistent storage and ReasoningBank integration. Enables agents to remember conversations, learn from interactions, and maintain context across sessions.
6
7	Performance: 150x-12,500x faster than traditional solutions with 100% backward compatibility.
8
9	## Prerequisites
10
11	- Node.js 18+
12	- AgentDB v1.0.7+ (via agentic-flow or standalone)
13	- Understanding of agent architectures
14
15	## Quick Start with CLI
16
17	### Initialize AgentDB
18
19	```bash
20	# Initialize vector database
21	npx agentdb@latest init ./agents.db
22
23	# Or with custom dimensions
24	npx agentdb@latest init ./agents.db --dimension 768
25
26	# Use preset configurations
27	npx agentdb@latest init ./agents.db --preset large
28
29	# In-memory database for testing
30	npx agentdb@latest init ./memory.db --in-memory
31	```
32
33	### Start MCP Server for Claude Code
34
35	```bash
36	# Start MCP server (integrates with Claude Code)
37	npx agentdb@latest mcp
38
39	# Add to Claude Code (one-time setup)
40	claude mcp add agentdb npx agentdb@latest mcp
41	```
42
43	### Create Learning Plugin
44
45	```bash
46	# Interactive plugin wizard
47	npx agentdb@latest create-plugin
48
49	# Use template directly
50	npx agentdb@latest create-plugin -t decision-transformer -n my-agent
51
52	# Available templates:
53	# - decision-transformer (sequence modeling RL)
54	# - q-learning (value-based learning)
55	# - sarsa (on-policy TD learning)
56	# - actor-critic (policy gradient)
57	# - curiosity-driven (exploration-based)
58	```
59
60	## Quick Start with API
61
62	```typescript
63	import { createAgentDBAdapter } from 'agentic-flow/reasoningbank';
64
65	// Initialize with default configuration
66	const adapter = await createAgentDBAdapter({
67	dbPath: '.agentdb/reasoningbank.db',
68	enableLearning: true, // Enable learning plugins
69	enableReasoning: true, // Enable reasoning agents
70	quantizationType: 'scalar', // binary \| scalar \| product \| none
71	cacheSize: 1000, // In-memory cache
72	});
73
74	// Store interaction memory
75	const patternId = await adapter.insertPattern({
76	id: '',
77	type: 'pattern',
78	domain: 'conversation',
79	pattern_data: JSON.stringify({
80	embedding: await computeEmbedding('What is the capital of France?'),
81	pattern: {
82	user: 'What is the capital of France?',
83	assistant: 'The capital of France is Paris.',
84	timestamp: Date.now()
85	}
86	}),
87	confidence: 0.95,
88	usage_count: 1,
89	success_count: 1,
90	created_at: Date.now(),
91	last_used: Date.now(),
92	});
93
94	// Retrieve context with reasoning
95	const context = await adapter.retrieveWithReasoning(queryEmbedding, {
96	domain: 'conversation',
97	k: 10,
98	useMMR: true, // Maximal Marginal Relevance
99	synthesizeContext: true, // Generate rich context
100	});
101	```
102
103	## Memory Patterns
104
105	### 1. Session Memory
106	```typescript
107	class SessionMemory {
108	async storeMessage(role: string, content: string) {
109	return await db.storeMemory({
110	sessionId: this.sessionId,
111	role,
112	content,
113	timestamp: Date.now()
114	});
115	}
116
117	async getSessionHistory(limit = 20) {
118	return await db.query({
119	filters: { sessionId: this.sessionId },
120	orderBy: 'timestamp',
121	limit
122	});
123	}
124	}
125	```
126
127	### 2. Long-Term Memory
128	```typescript
129	// Store important facts
130	await db.storeFact({
131	category: 'user_preference',
132	key: 'language',
133	value: 'English',
134	confidence: 1.0,
135	source: 'explicit'
136	});
137
138	// Retrieve facts
139	const prefs = await db.getFacts({
140	category: 'user_preference'
141	});
142	```
143
144	### 3. Pattern Learning
145	```typescript
146	// Learn from successful interactions
147	await db.storePattern({
148	trigger: 'user_asks_time',
149	response: 'provide_formatted_time',
150	success: true,
151	context: { timezone: 'UTC' }
152	});
153
154	// Apply learned patterns
155	const pattern = await db.matchPattern(currentContext);
156	```
157
158	## Advanced Patterns
159
160	### Hierarchical Memory
161	```typescript
162	// Organize memory in hierarchy
163	await memory.organize({
164	immediate: recentMessages, // Last 10 messages
165	shortTerm: sessionContext, // Current session
166	longTerm: importantFacts, // Persistent facts
167	semantic: embeddedKnowledge // Vector search
168	});
169	```
170
171	### Memory Consolidation
172	```typescript
173	// Periodically consolidate memories
174	await memory.consolidate({
175	strategy: 'importance', // Keep important memories
176	maxSize: 10000, // Size limit
177	minScore: 0.5 // Relevance threshold
178	});
179	```
180
181	## CLI Operations
182
183	### Query Database
184
185	```bash
186	# Query with vector embedding
187	npx agentdb@latest query ./agents.db "[0.1,0.2,0.3,...]"
188
189	# Top-k results
190	npx agentdb@latest query ./agents.db "[0.1,0.2,0.3]" -k 10
191
192	# With similarity threshold
193	npx agentdb@latest query ./agents.db "0.1 0.2 0.3" -t 0.75
194
195	# JSON output
196	npx agentdb@latest query ./agents.db "[...]" -

People who liked this also viewed

6 picks

Type

agentdb-memory-patterns

301stars24forks

View on GitHub

How to install agentdb-memory-patterns?

spencermarx/open-code-review/agentdb-memory-patterns

$npx -y skills add spencermarx/open-code-review --skill agentdb-memory-patterns

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# AgentDB Memory Patterns
2
3	## What This Skill Does
4
5	Provides memory management patterns for AI agents using AgentDB's persistent storage and ReasoningBank integration. Enables agents to remember conversations, learn from interactions, and maintain context across sessions.
6
7	Performance: 150x-12,500x faster than traditional solutions with 100% backward compatibility.
8
9	## Prerequisites
10
11	- Node.js 18+
12	- AgentDB v1.0.7+ (via agentic-flow or standalone)
13	- Understanding of agent architectures
14
15	## Quick Start with CLI
16
17	### Initialize AgentDB
18
19	```bash
20	# Initialize vector database
21	npx agentdb@latest init ./agents.db
22
23	# Or with custom dimensions
24	npx agentdb@latest init ./agents.db --dimension 768
25
26	# Use preset configurations
27	npx agentdb@latest init ./agents.db --preset large
28
29	# In-memory database for testing
30	npx agentdb@latest init ./memory.db --in-memory
31	```
32
33	### Start MCP Server for Claude Code
34
35	```bash
36	# Start MCP server (integrates with Claude Code)
37	npx agentdb@latest mcp
38
39	# Add to Claude Code (one-time setup)
40	claude mcp add agentdb npx agentdb@latest mcp
41	```
42
43	### Create Learning Plugin
44
45	```bash
46	# Interactive plugin wizard
47	npx agentdb@latest create-plugin
48
49	# Use template directly
50	npx agentdb@latest create-plugin -t decision-transformer -n my-agent
51
52	# Available templates:
53	# - decision-transformer (sequence modeling RL)
54	# - q-learning (value-based learning)
55	# - sarsa (on-policy TD learning)
56	# - actor-critic (policy gradient)
57	# - curiosity-driven (exploration-based)
58	```
59
60	## Quick Start with API
61
62	```typescript
63	import { createAgentDBAdapter } from 'agentic-flow/reasoningbank';
64
65	// Initialize with default configuration
66	const adapter = await createAgentDBAdapter({
67	dbPath: '.agentdb/reasoningbank.db',
68	enableLearning: true, // Enable learning plugins
69	enableReasoning: true, // Enable reasoning agents
70	quantizationType: 'scalar', // binary \| scalar \| product \| none
71	cacheSize: 1000, // In-memory cache
72	});
73
74	// Store interaction memory
75	const patternId = await adapter.insertPattern({
76	id: '',
77	type: 'pattern',
78	domain: 'conversation',
79	pattern_data: JSON.stringify({
80	embedding: await computeEmbedding('What is the capital of France?'),
81	pattern: {
82	user: 'What is the capital of France?',
83	assistant: 'The capital of France is Paris.',
84	timestamp: Date.now()
85	}
86	}),
87	confidence: 0.95,
88	usage_count: 1,
89	success_count: 1,
90	created_at: Date.now(),
91	last_used: Date.now(),
92	});
93
94	// Retrieve context with reasoning
95	const context = await adapter.retrieveWithReasoning(queryEmbedding, {
96	domain: 'conversation',
97	k: 10,
98	useMMR: true, // Maximal Marginal Relevance
99	synthesizeContext: true, // Generate rich context
100	});
101	```
102
103	## Memory Patterns
104
105	### 1. Session Memory
106	```typescript
107	class SessionMemory {
108	async storeMessage(role: string, content: string) {
109	return await db.storeMemory({
110	sessionId: this.sessionId,
111	role,
112	content,
113	timestamp: Date.now()
114	});
115	}
116
117	async getSessionHistory(limit = 20) {
118	return await db.query({
119	filters: { sessionId: this.sessionId },
120	orderBy: 'timestamp',
121	limit
122	});
123	}
124	}
125	```
126
127	### 2. Long-Term Memory
128	```typescript
129	// Store important facts
130	await db.storeFact({
131	category: 'user_preference',
132	key: 'language',
133	value: 'English',
134	confidence: 1.0,
135	source: 'explicit'
136	});
137
138	// Retrieve facts
139	const prefs = await db.getFacts({
140	category: 'user_preference'
141	});
142	```
143
144	### 3. Pattern Learning
145	```typescript
146	// Learn from successful interactions
147	await db.storePattern({
148	trigger: 'user_asks_time',
149	response: 'provide_formatted_time',
150	success: true,
151	context: { timezone: 'UTC' }
152	});
153
154	// Apply learned patterns
155	const pattern = await db.matchPattern(currentContext);
156	```
157
158	## Advanced Patterns
159
160	### Hierarchical Memory
161	```typescript
162	// Organize memory in hierarchy
163	await memory.organize({
164	immediate: recentMessages, // Last 10 messages
165	shortTerm: sessionContext, // Current session
166	longTerm: importantFacts, // Persistent facts
167	semantic: embeddedKnowledge // Vector search
168	});
169	```
170
171	### Memory Consolidation
172	```typescript
173	// Periodically consolidate memories
174	await memory.consolidate({
175	strategy: 'importance', // Keep important memories
176	maxSize: 10000, // Size limit
177	minScore: 0.5 // Relevance threshold
178	});
179	```
180
181	## CLI Operations
182
183	### Query Database
184
185	```bash
186	# Query with vector embedding
187	npx agentdb@latest query ./agents.db "[0.1,0.2,0.3,...]"
188
189	# Top-k results
190	npx agentdb@latest query ./agents.db "[0.1,0.2,0.3]" -k 10
191
192	# With similarity threshold
193	npx agentdb@latest query ./agents.db "0.1 0.2 0.3" -t 0.75
194
195	# JSON output
196	npx agentdb@latest query ./agents.db "[...]" -

People who liked this also viewed

6 picks

Type