AgentMemory

AgentMemory is a persistent memory runtime that captures, recalls, and consolidates coding agent sessions.

AgentMemory is a persistent memory runtime designed for AI coding agents to survive across sessions without bloating the context window. It captures tool calls, prompts, and session stops via twelve auto-capture hooks and makes them accessible through MCP tools and REST endpoints. The system uses hybrid retrieval combining BM25, vector, and knowledge graph scores, running locally as a single Node process. It supports auto-consolidation of observations into semantic memories when provided with an LLM provider key. Native plugins are available for tools like Claude Code, Cursor, Copilot CLI, and others.

AgentMemory at a glance

Best for: Software engineer
Runs on: Web, API, CLI, self-hosted
Pricing: Free
Released: February 2026
Category: AI coding tool
Website: agent-memory.dev

Key features

Capture every session start, prompt, tool call, and stop
Hybrid recall with BM25, vector, and knowledge graph scores
Auto-consolidation of observations into semantic memories
Native Model Context Protocol surface with multiple tools
HTTP first REST endpoints under agentmemory path
Knowledge graph extraction with temporal edges
Peer-to-peer sync between nodes over authenticated HTTPS
Obsidian export to mirror memories as markdown

Use it for

Retain project context across multiple coding agent sessions
Import and rehydrate past session transcripts from JSONL
Query and visualize knowledge graphs of entities and relations
Sync memories across peer-to-peer nodes

Alternatives to AgentMemory

Supermemory — Integrates long-term memory into AI coding agents with a single prompt.
MuscleMemory — It cuts input tokens, tool calls, and execution time while maintaining task success.
LongMemory — Provides immutable, temporal, and governed recall instead of simple vector retrieval.
claude-mem — It shares persistent, searchable agent memory across sessions and devices.
memU — It gives every connected agent a shared and inspectable memory layer.
Memori — Reduces LLM token usage and spend by over 97% with targeted recall.
Mem0 — It provides drop-in persistent memory for AI agents without changing pipelines.
ReMe — It enables local-first, file-native long-term memory sharing across multiple AI agents.
AI Memory — Shared memory for coding agents that works across machines and team members.
memsearch — It shares continuous context across Claude Code, Codex, and other AI coding agents.
MemOS — It delivers ultra-persistent memory, hybrid-retrieval, and token savings.
ProjectMem — Records project history locally to prevent AI from repeating failed work.

All AgentMemory alternatives