awesome-agent-memory
A structured, decision-grade evidence base on long-term memory for LLM agents, combining a 989-paper index, archived source pages, product notes, benchmark protocols and architecture maps.
awesome-agent-memory is more than a link list: it separates evidence about agent memory into five layers, papers with scholarly claims and reading notes, public behavior of memory products and platforms, evaluation protocols and vendor claims for benchmarks, a synthesis layer with taxonomy and architecture patterns, and a research-to-decision workflow that turns evidence into inputs for architecture decisions.
Coverage as of the README includes 989 scraped papers plus manual additions, 988 paper stubs, 534 archived PDFs, 40 memory-product notes with 39 page-archive snapshots, and 23 benchmark catalog entries. Direct evidence, maintainer inference, vendor self-claims, affiliated evaluations and independent reproductions are kept in separate buckets throughout.
It is available in Chinese and English, and licensed under Apache-2.0.