Show HN: An open Add/Search evaluation framework for agent memory

Category: ai-ml

Tags: agent-memory, evaluation, benchmark, llm, leaderboard, add-search-api

Score: 6.0/10 (Innovation: 6, Technical: 5, Documentation: 7, Utility: 6)

Agent Memory Leaderboard (AML) is an open evaluation framework and public leaderboard for long-term memory systems used by AI agents. By standardizing the Add/Search interface and controlling the downstream answer and scoring pipeline, it enables reproducible, apples-to-apples comparisons across memory implementations. The project is notable for its broad capability taxonomy covering textual and coding memory, plus a community challenge cycle with contributions from over twenty universities.

Target audience: AI/ML researchers, agent framework developers, LLM engineers, and teams building or benchmarking long-term memory for autonomous agents

Repository: https://agentmemoryleaderboard.ai/ · Python · 1180 stars

View on Hacker News