Skip to content

Cookbooks

Cookbooks are step-by-step guides that demonstrate how to use HEBBS in real-world scenarios. Each cookbook includes complete, runnable code and explains the design decisions behind the implementation.

Build foundational skills with HEBBS:

  • Your First Memory Agent — connect to HEBBS and build a simple memory-augmented agent from scratch.
  • Multi-Strategy Recall — compare all four recall strategies (similarity, temporal, causal, analogical) on the same dataset.
  • Entity-Scoped Memory — partition memories by customer, user, or agent and prove isolation.

Complete applications demonstrating HEBBS in production scenarios:

  • Voice Sales Agent — build an AI sales agent with cross-session memory for discovery calls and objection handling.
  • Customer Support Agent — support agent with ticket history, temporal recall, and GDPR-compliant data erasure.
  • Research Assistant — cross-domain recall and synthesis for research workflows using analogical recall and reflection.

Reusable patterns for common requirements:

  • GDPR Compliance — implement data erasure with forget() and verify complete deletion.
  • Real-time Subscribe — surface relevant memories in real-time during conversations.
  • Background Learning — use the reflection pipeline to consolidate memories into institutional knowledge.
  • Causal Chains — build and query causal relationships between memories.

Operational guides for running HEBBS:

  • Monitoring Stack — set up HEBBS with Prometheus and Grafana using docker-compose.

All cookbooks assume:

  • A running HEBBS server (see Deployment)
  • Python: The HEBBS Python SDK installed (pip install hebbs) with Python 3.10+
  • TypeScript: The HEBBS TypeScript SDK installed (npm install @hebbs/sdk) with Node.js 18+

Most examples show both Python and TypeScript. Pick whichever fits your stack.