reasoningbank-agentdb

Name: reasoningbank-agentdb
Author: spencermarx

301stars24forks

How to install reasoningbank-agentdb?

spencermarx/open-code-review/reasoningbank-agentdb

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

Installs into the current project.

Summary

Implement ReasoningBank adaptive learning with AgentDB's 150x faster vector database. Includes trajectory tracking, verdict judgment, memory distillation, and pattern recognition. Use when building self-learning agents, optimizing decision-making, or implementing experience repla

Files · 1

View on GitHub

SKILL.md

1	# ReasoningBank with AgentDB
2
3	## What This Skill Does
4
5	Provides ReasoningBank adaptive learning patterns using AgentDB's high-performance backend (150x-12,500x faster). Enables agents to learn from experiences, judge outcomes, distill memories, and improve decision-making over time with 100% backward compatibility.
6
7	Performance: 150x faster pattern retrieval, 500x faster batch operations, <1ms memory access.
8
9	## Prerequisites
10
11	- Node.js 18+
12	- AgentDB v1.0.7+ (via agentic-flow)
13	- Understanding of reinforcement learning concepts (optional)
14
15	---
16
17	## Quick Start with CLI
18
19	### Initialize ReasoningBank Database
20
21	```bash
22	# Initialize AgentDB for ReasoningBank
23	npx agentdb@latest init ./.agentdb/reasoningbank.db --dimension 1536
24
25	# Start MCP server for Claude Code integration
26	npx agentdb@latest mcp
27	claude mcp add agentdb npx agentdb@latest mcp
28	```
29
30	### Migrate from Legacy ReasoningBank
31
32	```bash
33	# Automatic migration with validation
34	npx agentdb@latest migrate --source .swarm/memory.db
35
36	# Verify migration
37	npx agentdb@latest stats ./.agentdb/reasoningbank.db
38	```
39
40	---
41
42	## Quick Start with API
43
44	```typescript
45	import { createAgentDBAdapter, computeEmbedding } from 'agentic-flow/reasoningbank';
46
47	// Initialize ReasoningBank with AgentDB
48	const rb = await createAgentDBAdapter({
49	dbPath: '.agentdb/reasoningbank.db',
50	enableLearning: true, // Enable learning plugins
51	enableReasoning: true, // Enable reasoning agents
52	cacheSize: 1000, // 1000 pattern cache
53	});
54
55	// Store successful experience
56	const query = "How to optimize database queries?";
57	const embedding = await computeEmbedding(query);
58
59	await rb.insertPattern({
60	id: '',
61	type: 'experience',
62	domain: 'database-optimization',
63	pattern_data: JSON.stringify({
64	embedding,
65	pattern: {
66	query,
67	approach: 'indexing + query optimization',
68	outcome: 'success',
69	metrics: { latency_reduction: 0.85 }
70	}
71	}),
72	confidence: 0.95,
73	usage_count: 1,
74	success_count: 1,
75	created_at: Date.now(),
76	last_used: Date.now(),
77	});
78
79	// Retrieve similar experiences with reasoning
80	const result = await rb.retrieveWithReasoning(embedding, {
81	domain: 'database-optimization',
82	k: 5,
83	useMMR: true, // Diverse results
84	synthesizeContext: true, // Rich context synthesis
85	});
86
87	console.log('Memories:', result.memories);
88	console.log('Context:', result.context);
89	console.log('Patterns:', result.patterns);
90	```
91
92	---
93
94	## Core ReasoningBank Concepts
95
96	### 1. Trajectory Tracking
97
98	Track agent execution paths and outcomes:
99
100	```typescript
101	// Record trajectory (sequence of actions)
102	const trajectory = {
103	task: 'optimize-api-endpoint',
104	steps: [
105	{ action: 'analyze-bottleneck', result: 'found N+1 query' },
106	{ action: 'add-eager-loading', result: 'reduced queries' },
107	{ action: 'add-caching', result: 'improved latency' }
108	],
109	outcome: 'success',
110	metrics: { latency_before: 2500, latency_after: 150 }
111	};
112
113	const embedding = await computeEmbedding(JSON.stringify(trajectory));
114
115	await rb.insertPattern({
116	id: '',
117	type: 'trajectory',
118	domain: 'api-optimization',
119	pattern_data: JSON.stringify({ embedding, pattern: trajectory }),
120	confidence: 0.9,
121	usage_count: 1,
122	success_count: 1,
123	created_at: Date.now(),
124	last_used: Date.now(),
125	});
126	```
127
128	### 2. Verdict Judgment
129
130	Judge whether a trajectory was successful:
131
132	```typescript
133	// Retrieve similar past trajectories
134	const similar = await rb.retrieveWithReasoning(queryEmbedding, {
135	domain: 'api-optimization',
136	k: 10,
137	});
138
139	// Judge based on similarity to successful patterns
140	const verdict = similar.memories.filter(m =>
141	m.pattern.outcome === 'success' &&
142	m.similarity > 0.8
143	).length > 5 ? 'likely_success' : 'needs_review';
144
145	console.log('Verdict:', verdict);
146	console.log('Confidence:', similar.memories[0]?.similarity \|\| 0);
147	```
148
149	### 3. Memory Distillation
150
151	Consolidate similar experiences into patterns:
152
153	```typescript
154	// Get all experiences in domain
155	const experiences = await rb.retrieveWithReasoning(embedding, {
156	domain: 'api-optimization',
157	k: 100,
158	optimizeMemory: true, // Automatic consolidation
159	});
160
161	// Distill into high-level pattern
162	const distilledPattern = {
163	domain: 'api-optimization',
164	pattern: 'For N+1 queries: add eager loading, then cache',
165	success_rate: 0.92,
166	sample_size: experiences.memories.length,
167	confidence: 0.95
168	};
169
170	await rb.insertPattern({
171	id: '',
172	type: 'distilled-pattern',
173	domain: 'api-optimization',
174	pattern_data: JSON.stringify({
175	embedding: await computeEmbedding(JSON.stringify(distilledPattern)),
176	pattern: distilledPattern
177	}),
178	confidence: 0.95,
179	usage_count: 0,
180	success_count: 0,
181	created_at: Da

People who liked this also viewed

6 picks

Type

reasoningbank-agentdb

301stars24forks

View on GitHub

How to install reasoningbank-agentdb?

spencermarx/open-code-review/reasoningbank-agentdb

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

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# ReasoningBank with AgentDB
2
3	## What This Skill Does
4
5	Provides ReasoningBank adaptive learning patterns using AgentDB's high-performance backend (150x-12,500x faster). Enables agents to learn from experiences, judge outcomes, distill memories, and improve decision-making over time with 100% backward compatibility.
6
7	Performance: 150x faster pattern retrieval, 500x faster batch operations, <1ms memory access.
8
9	## Prerequisites
10
11	- Node.js 18+
12	- AgentDB v1.0.7+ (via agentic-flow)
13	- Understanding of reinforcement learning concepts (optional)
14
15	---
16
17	## Quick Start with CLI
18
19	### Initialize ReasoningBank Database
20
21	```bash
22	# Initialize AgentDB for ReasoningBank
23	npx agentdb@latest init ./.agentdb/reasoningbank.db --dimension 1536
24
25	# Start MCP server for Claude Code integration
26	npx agentdb@latest mcp
27	claude mcp add agentdb npx agentdb@latest mcp
28	```
29
30	### Migrate from Legacy ReasoningBank
31
32	```bash
33	# Automatic migration with validation
34	npx agentdb@latest migrate --source .swarm/memory.db
35
36	# Verify migration
37	npx agentdb@latest stats ./.agentdb/reasoningbank.db
38	```
39
40	---
41
42	## Quick Start with API
43
44	```typescript
45	import { createAgentDBAdapter, computeEmbedding } from 'agentic-flow/reasoningbank';
46
47	// Initialize ReasoningBank with AgentDB
48	const rb = await createAgentDBAdapter({
49	dbPath: '.agentdb/reasoningbank.db',
50	enableLearning: true, // Enable learning plugins
51	enableReasoning: true, // Enable reasoning agents
52	cacheSize: 1000, // 1000 pattern cache
53	});
54
55	// Store successful experience
56	const query = "How to optimize database queries?";
57	const embedding = await computeEmbedding(query);
58
59	await rb.insertPattern({
60	id: '',
61	type: 'experience',
62	domain: 'database-optimization',
63	pattern_data: JSON.stringify({
64	embedding,
65	pattern: {
66	query,
67	approach: 'indexing + query optimization',
68	outcome: 'success',
69	metrics: { latency_reduction: 0.85 }
70	}
71	}),
72	confidence: 0.95,
73	usage_count: 1,
74	success_count: 1,
75	created_at: Date.now(),
76	last_used: Date.now(),
77	});
78
79	// Retrieve similar experiences with reasoning
80	const result = await rb.retrieveWithReasoning(embedding, {
81	domain: 'database-optimization',
82	k: 5,
83	useMMR: true, // Diverse results
84	synthesizeContext: true, // Rich context synthesis
85	});
86
87	console.log('Memories:', result.memories);
88	console.log('Context:', result.context);
89	console.log('Patterns:', result.patterns);
90	```
91
92	---
93
94	## Core ReasoningBank Concepts
95
96	### 1. Trajectory Tracking
97
98	Track agent execution paths and outcomes:
99
100	```typescript
101	// Record trajectory (sequence of actions)
102	const trajectory = {
103	task: 'optimize-api-endpoint',
104	steps: [
105	{ action: 'analyze-bottleneck', result: 'found N+1 query' },
106	{ action: 'add-eager-loading', result: 'reduced queries' },
107	{ action: 'add-caching', result: 'improved latency' }
108	],
109	outcome: 'success',
110	metrics: { latency_before: 2500, latency_after: 150 }
111	};
112
113	const embedding = await computeEmbedding(JSON.stringify(trajectory));
114
115	await rb.insertPattern({
116	id: '',
117	type: 'trajectory',
118	domain: 'api-optimization',
119	pattern_data: JSON.stringify({ embedding, pattern: trajectory }),
120	confidence: 0.9,
121	usage_count: 1,
122	success_count: 1,
123	created_at: Date.now(),
124	last_used: Date.now(),
125	});
126	```
127
128	### 2. Verdict Judgment
129
130	Judge whether a trajectory was successful:
131
132	```typescript
133	// Retrieve similar past trajectories
134	const similar = await rb.retrieveWithReasoning(queryEmbedding, {
135	domain: 'api-optimization',
136	k: 10,
137	});
138
139	// Judge based on similarity to successful patterns
140	const verdict = similar.memories.filter(m =>
141	m.pattern.outcome === 'success' &&
142	m.similarity > 0.8
143	).length > 5 ? 'likely_success' : 'needs_review';
144
145	console.log('Verdict:', verdict);
146	console.log('Confidence:', similar.memories[0]?.similarity \|\| 0);
147	```
148
149	### 3. Memory Distillation
150
151	Consolidate similar experiences into patterns:
152
153	```typescript
154	// Get all experiences in domain
155	const experiences = await rb.retrieveWithReasoning(embedding, {
156	domain: 'api-optimization',
157	k: 100,
158	optimizeMemory: true, // Automatic consolidation
159	});
160
161	// Distill into high-level pattern
162	const distilledPattern = {
163	domain: 'api-optimization',
164	pattern: 'For N+1 queries: add eager loading, then cache',
165	success_rate: 0.92,
166	sample_size: experiences.memories.length,
167	confidence: 0.95
168	};
169
170	await rb.insertPattern({
171	id: '',
172	type: 'distilled-pattern',
173	domain: 'api-optimization',
174	pattern_data: JSON.stringify({
175	embedding: await computeEmbedding(JSON.stringify(distilledPattern)),
176	pattern: distilledPattern
177	}),
178	confidence: 0.95,
179	usage_count: 0,
180	success_count: 0,
181	created_at: Da

People who liked this also viewed

6 picks

Type