RE
ReasoningBank with AgentDB
Implement adaptive learning for AI agents using AgentDB's vector database backend for trajectory tracking, verdict judgment, memory distillation, and pattern recognition. Use when building self-learning agents, implementing reinforcement learning patterns, or migrating from legacy ReasoningBank to A
Install
mkdir -p .claude/skills/reasoningbank-with-agentdb-wiktorskrabel89-byte && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14503" && unzip -o skill.zip -d .claude/skills/reasoningbank-with-agentdb-wiktorskrabel89-byte && rm skill.zipInstalls to .claude/skills/reasoningbank-with-agentdb-wiktorskrabel89-byte
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Implement adaptive learning for AI agents using AgentDB's vector database backend for trajectory tracking, verdict judgment, memory distillation, and pattern recognition. Use when building self-learning agents, implementing reinforcement learning patterns, or migrating from legacy ReasoningBank to AgentDB.307 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
About this skill
ReasoningBank with AgentDB
Implement adaptive learning for AI agents with 150x faster pattern retrieval, 500x faster batch operations, and <1ms memory access.
Prerequisites
- Node.js 18+
- AgentDB v1.0.7+
- agentic-flow installed
Quick Start
import { createAgentDBAdapter } from 'agentic-flow/reasoningbank';
const adapter = await createAgentDBAdapter({
dbPath: '.agentdb/reasoningbank.db',
enableLearning: true,
enableReasoning: true,
});
// Start trajectory tracking
await adapter.startTrajectory({
task: 'Implement user authentication',
context: { language: 'typescript' }
});
// Record action and outcome
await adapter.recordAction({
action: 'create_jwt_middleware',
result: 'success',
reward: 0.9
});
// Finalize trajectory
await adapter.finalizeTrajectory({ success: true, score: 0.92 });
Four Reasoning Modules
- PatternMatcher - Find similar successful patterns using diversity-aware HNSW retrieval
- ContextSynthesizer - Generate coherent narratives from multiple memory sources
- MemoryOptimizer - Automatically consolidate and remove low-quality patterns
- ExperienceCurator - Filter memories by quality thresholds
Backward Compatibility
Full compatibility with legacy ReasoningBank APIs - AgentDB backend used automatically.
Migration from Legacy
# Migrate existing ReasoningBank data
npx agentdb@latest migrate --source .swarm/memory.db --dest .agentdb/reasoningbank.db
Performance
- Pattern retrieval: 150x faster
- Batch operations: 500x faster
- Memory access: <1ms