Frequently asked questions
Octopodas is an open source infrastructure layer designed to provide AI agents with persistent memory, automatic loop detection, and full observability. It addresses the common issue where agents forget everything between sessions by implementing a remember/recall API that survives restarts, crashes, and deployments. Additionally, it features a loop detection system that monitors multiple signals to prevent agents from repeating actions unnecessarily, which can waste API budgets. Users benefit from a built-in dashboard that displays each agent's health score, memory history, and an audit trail of decisions made by the agent. Octopodas integrates seamlessly with popular frameworks like LangChain and OpenAI Agents SDK, and it can run locally or connect to the cloud easily. It is available for free under the MIT license.
