danielcase-man/engram

Engram is a utility project targeting the very recent 'Claude Code' release from Anthropic. While the problem it solves (long-term memory for agents) is critical, the project currently has zero stars, forks, or community traction. From a competitive standpoint, this is a classic 'feature-gap' project. It fills a void in a newly released frontier product (Claude Code's lack of native multi-session memory) that the platform provider is almost certain to fill themselves within months. Defensibility is minimal as the logic for saving agent history to a local store or vector DB is standard industry practice, already implemented by more mature frameworks like MemGPT (now Letta), LangGraph, or CrewAI. Anthropic has a high incentive to build this directly into the Claude Code binary to improve user retention and agent performance, making this project's survival window very short. The 'displacement horizon' is set to 6 months because that is the likely timeframe for the next major iteration of frontier-lab coding agents to include native state management.