agentdb-memory-patterns

name

agentdb-memory-patterns

description

Implement persistent memory patterns for AI agents using AgentDB — session memory, long-term storage, pattern learning, hierarchical memory, consolidation, ReasoningBank migration. Use when building stateful agents, chat systems, or intelligent assistants. NOT for distributed multi-DB/QUIC sync or RL algorithms (use agentdb-advanced), vector-search perf tuning (use agentdb-vector-search), or non-persistent session state.

AgentDB Memory Patterns

For AgentDB architecture and performance overview, see AgentDB Overview.

What This Skill Covers

Memory management patterns for AI agents: session memory, long-term storage, pattern learning, hierarchical memory organisation, memory consolidation, and ReasoningBank migration. For CLI setup, API initialisation, and common operations, see the AgentDB Overview.

Additional Prerequisites: Understanding of agent architectures.

When Not To Use

• For distributed multi-database setups or QUIC sync -- use the agentdb-advanced skill instead
• For reinforcement learning algorithms -- use the agentdb-advanced skill (RL Plugins section) instead
• For pure vector search performance tuning (quantisation, HNSW) -- use the agentdb-vector-search skill instead
• For non-persistent, session-only state that does not need vector search -- standard in-memory data structures suffice
• For unified development and quality engineering workflows -- use the build-with-quality skill which includes memory management

Pattern Reference

Full code examples live in references/pattern-examples.md:

• Session Memory — per-session message store with time-ordered history.
• Long-Term Memory — categorised facts with confidence and source.
• Pattern Learning — record successful trigger→response pairs and match them against live context.
• Hierarchical Memory — organise across immediate / short-term / long-term / semantic tiers.
• Memory Consolidation — periodic importance-based pruning to a size/score threshold.
• ReasoningBank Integration — migrate a legacy .swarm/memory.db, train the learning model, and retrieve with reasoning.

For learning plugins and RL algorithms, see the agentdb-advanced skill (RL Plugins section). For performance tuning and quantization, see the agentdb-vector-search skill.

KHIVE-Informed Enhancements (v2.0 Roadmap)

Proposed enhancements from KHIVE v2 comparison — orient cold-start dashboard, importance-weighted storage, entity-relationship graph, task dependency DAG, tag retrieval path, and URI/URN addressing. Full proposal with schemas and code in references/khive-v2-roadmap.md; underlying analysis in KHIVE-LEARNINGS.md.

Troubleshooting

Issue: Memory growing too large

# Check database size
npx agentdb@latest stats ./agents.db

# Enable quantization
# Use 'binary' (32x smaller) or 'scalar' (4x smaller)

Issue: Slow search performance

# Enable HNSW indexing and caching
# Results: <100µs search time

Issue: Migration from legacy ReasoningBank

# Automatic migration with validation
npx agentdb@latest migrate --source .swarm/memory.db

Learn More

For performance benchmarks, see AgentDB Overview. For general links, see AgentDB Overview.