Hexabot v3: When Chatbots Need Contracts and Memory

What it does Hexabot v3 is an automation platform that runs agentic workflows and conversational channels inside a single TypeScript runtime. You define workflows in YAML with typed contracts, wire actions together through a binding system, and expose the results across chat channels. Under the hood it uses Zod for validation and TypeORM to talk to either SQLite or Postgres.

The interesting bit The project treats memory as a first-class citizen rather than an afterthought, giving workflows explicit memory definitions and runtime integration. It also speaks MCP (Model Context Protocol), positioning itself as a context-aware hub rather than a siloed bot framework.

Key highlights

• YAML workflow definitions with typed runtime contracts and schema-validated action I/O.
• Reusable binding system that separates capability configuration from task logic.
• Explicit memory definitions with runtime memory integration.
• MCP integration points for tool and context interoperability.
• Multi-channel continuity for cross-surface conversations.

Caveats

• The CLI requires an interactive terminal (TTY) for initial project creation, so headless CI setup is not supported out of the box.
• The tagline advertises RAG, but the core capabilities never explain how retrieval or vector storage is wired in.

Verdict If you are building chatbots that need structured, long-running workflows with explicit state and cross-channel persistence, Hexabot v3 is worth a look. If you want a lightweight, bring-your-own-LLM script runner, this is probably more platform than you need.