The best agent memory systems, according to hackers
20 agent memory systems recommended in Hacker News comments over the past year, ranked by how many separate people recommend each one.
All agent memory systems
- EngrimOpen-source local memory engine for AI agents, built on SQLite, with local embeddings and no API key or external service needed. Critics note it is still limited by effective context size.21recommend
- Total Recall for Claude CodeRecommended as a write-gated memory layer for Claude Code that keeps only entries worth saving, so past decisions and conversations stay searchable.12recommend
- OKF Agent MemoryGit-native memory for AI coding agents, a zero-dependency Go binary that preserves context across handoffs. Critics say it adds little beyond local markdown files.9recommend
- MemPalacePeople recommend it as a locally managed memory layer for AI agents that retrieves past conversations well. Critics say independent end-to-end tests scored far below its retrieval benchmarks.9recommend
- GrovPeople recommend it for cutting exploration tokens and recalling past sessions for Claude Code. One user says it is rough around the edges, and it needs an API key.8recommend
- claude-memRecommended as an open-source Claude Code memory plugin that recalls past prompts and replies without many tokens. One caveat: it only records from install forward.8recommend
- LettaFree open-source agent framework with selective memory retrieval and long-context handling. Critics point to heavy dependencies and embedding API costs.8recommend
- Hippo AI agent memoryRecommended for modeling memory consolidation and decay for AI agents rather than a plain vector store. Critics question whether decay is needed.8recommend
- ContextifyMac app that backs up and summarizes Claude Code and Codex conversations into a database your AI can read.6recommend
- Mem0Recommended as a middle layer for selective, real-time memory retrieval and updates in AI agents. Critics say it depends heavily on LLM and embedding API calls, which adds cost.6recommend
- Claude BrainA memory project for Claude Code that people say works well when CLAUDE.md tells the model when to save and search memories. Critics say it is a thin wrapper around Memvid, not an Anthropic product.6recommend
- JevmemRepo-owned decision log that keeps project memory across Claude Code sessions without manual prompting. Caveat: sends conversation and memory data to TypeSafe AI, a young vendor.6recommend
- Lossless-memoryPeople recommend it as a personal AI memory that keeps full records, with useful versioned chronology. Caveat: relative time parsing is hardcoded for Japanese.6recommend
- HindsightPeople recommend Hindsight as a memory layer for AI agents that can cut token use over time and replace hand-kept memory files. The caveat is that stored memories were sometimes ignored by agents after a while.5recommend
- DriftOSFree software that keeps multi-topic conversations and AI coding sessions organized in plain markdown and git, with no SaaS or lock-in.5recommend
- Funes by Hugging FacePlain-text memory for AI coding agents, favored for conversation search over vector search. Critics report a first indexing pass covered only 189 of 19k sessions.5recommend
- MemoryLanePeople recommend MemoryLane for recording screen activity and making it searchable, so Claude can recall past work and suggest automations.5recommend
- GBrainPeople recommend GBrain as provider-agnostic AI memory and a useful starting point for individuals. Critics say it needs technical knowledge and time to configure.5recommend
- MCP MemoryLightweight memory server for AI agents, built on SQLite FTS5 for fast search, and usable from any MCP client that lacks a filesystem, such as claude.ai.5recommend
- LCM: Lossless Context ManagementRecommended as a mechanistic way to manage LLM context with stronger guarantees when deployed in production.5recommend