Agent-Memory vs agentmemory
Persistent memory for AI coding agents compared
Agent-Memory and agentmemory both capture and retrieve project context across AI coding sessions, with shared access through agent integrations, MCP, and REST. Agent-Memory highlights memory search and inspection, while agentmemory details its search methods, memory tiers, and integrations for a range of coding agents.
Agent-Memory: Persistent Memory for AI Coding Agents on iii
Agent-Memory adds persistent, searchable memory to coding agents through the iii engine, sharing context across Claude Code, Cursor, Gemini CLI, Codex CLI, and other MCP or HTTP clients. It captures session activity and retrieves relevant context for later work.
agentmemory: Persistent Memory for AI Coding Agents
agentmemory gives coding agents persistent, searchable memory across sessions and tools. It captures work through agent integrations and exposes shared memory through MCP and REST, with a local-first runtime built on iii-engine.
| Agent-Memory | agentmemory | |
|---|---|---|
| Language | — | TypeScript |
| License | Apache-2.0 | Apache-2.0 |
| Stars | 9 | 29.1k |
| Forks | 4 | 2.5k |
| Last analyzed | Nov 1, 2026 | Oct 3, 2026 |
Key differences
- Agent-Memory describes keyword, vector, and knowledge-graph search; agentmemory specifies BM25 search with optional vector embeddings and graph-based matching.
- Agent-Memory describes memory compression, consolidation, and expiry; agentmemory describes working, episodic, semantic, and procedural memory tiers.
- Agent-Memory includes a web viewer for sessions, observations, memories, and graph data; agentmemory offers a real-time viewer and session replay.
- Agent-Memory is built on the iii engine; agentmemory specifies iii-engine version 0.22.1 and gives native Windows setup options.
- Both use the Apache-2.0 license and are not archived. The supplied metadata lists 6 stars and 1 fork for Agent-Memory, versus 29.1k stars and 2.5k forks for agentmemory.
- Agent-Memory’s considerations note that LLM-backed compression is disabled by default; agentmemory’s note that keyless setup supports BM25 but disables vector embeddings.
Choose Agent-Memory if you…
- want to inspect sessions, observations, memories, and graph data in a web viewer.
- need memory search that combines keyword, vector, and knowledge-graph signals.
- prefer the project’s described compression, consolidation, and expiry mechanisms.
Choose agentmemory if you…
- want documented integrations and setup paths for a range of coding agents.
- need explicit BM25 search, with optional vector embeddings and graph-based matching.
- want to review session replay and the project’s benchmark and evaluation materials.
Compare the numbers side by side, or add more projects
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.