AgenticMemory: A Binary Graph Format for Persistent, Portable, and Navigable AI Agent Memory
Defines a portable binary graph format for persistent AI-agent memory.
Research corpus for AgenticMemory, AgenticVision, and AgenticCodebase. Each paper documents core architecture decisions, data formats, and MCP integration pathways.
Total Papers
6
Projects
3
Open Access
100%
Defines a portable binary graph format for persistent AI-agent memory.
Extends memory with cognitive query capabilities and expanded retrieval logic.
Connects persistent memory infrastructure to agent tooling through MCP.
Describes rapid web cartography and structured extraction for agent workflows.
Formalizes persistent visual memory workflows delivered through MCP interfaces.
Introduces semantic code compilation for navigable and predictive code intelligence.
Compile mapped websites into typed APIs so agents import domains as libraries instead of browsing page-by-page workflows.
Delta-sharing architecture where each mapped site can reduce duplicate remapping and give new agents warm-start context.
Historical delta stacks for pattern detection, trend modeling, predictive alerts, and web-state memory over time.