Agentmemory
Open-source persistent memory runtime for AI coding agents — zero external databases.
If you rely on Claude Code or any MCP agent and struggle with context loss, Agentmemory is the most complete open-source memory runtime we've tested—zero external deps, 12 auto-capture hooks, and benchmark-proven recall. Skip it if you need cloud management or RBAC; mem0 or Letta fit managed needs, but for self-hosted control, Agentmemory leads.
Verified 2d ago · liveness 76/100 · cite: rightaichoice.com/tools/agentmemory
- Developers using Claude Code who need persistent memory across sessions
- Teams working on large codebases where agents lose context
- Power users of Cursor, Copilot, and other MCP agents
- Anyone building long-running autonomous coding workflows
- Users who need a cloud-hosted, fully managed solution
- Teams requiring built-in multi-user permissions or RBAC
- Complete beginners unfamiliar with CLI tools and MCP protocols
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Agentmemory if you need a fully managed cloud solution with built-in multi-user permissions (RBAC) or if you're a complete beginner unfamiliar with CLI tools and MCP protocols.
No cloud hosting is provided; you must run and maintain your own server process, which may incur infrastructure costs if deployed on a VPS or cloud instance.
Agentmemory is completely free and open-source (Apache-2.0) with no paid tiers, making it ideal for individual developers and small teams who want a self-hosted memory runtime without subscription costs. Compared to managed options like mem0 or Letta, which typically charge per seat or usage, Agentmemory offers a zero-cost entry point, though you'll need to provide your own LLM API key for advanced features. This pricing fits developers who are comfortable running their own infrastructure and
In short
Agentmemory — Open-source persistent memory runtime for AI coding agents — zero external databases. Best for Developers using Claude Code who need persistent memory across sessions, Teams working on large codebases where agents lose context, Power users of Cursor, Copilot, and other MCP agents. Free to use.
What's new in Agentmemory
Checked 2 days agoAcross the latest 1 update: 1 feature update.
What people actually say about Agentmemory — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
60 mentions across 6 sources (Hacker News, YouTube, Product Hunt, Bluesky, GitHub, Lemmy) · researched Jul 6, 2026.
- +Free and open-source (MIT) with no external databases required.
- +Excellent benchmark: 95.2% recall on LongMemEval-S.
- +Triple-stream retrieval (BM25, vector, knowledge graph) is unique and effective.
- +Works with 10+ coding agents via MCP and REST.
- +Reduces token usage by ~92%, saving cost and context window.
- −Crashes under large datasets (370K+ observations reported).
- −Some early adopters retracted after production use.
- −Tier 1 cap may limit truly infinite memory scenarios.
- −Manual pruning needed for contradictory or stale memories.
- −No clear business model raises questions about future support.
- • No paid tier or cloud version currently; self-hosting costs (server, storage) are on you.
Viability Score
How well maintained and how widely used is Agentmemory? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: August 2026
How we score →Key Features
- 12 auto-hooks capture session start, prompt, tool call, and stop
- Hybrid recall: BM25 + vector + knowledge graph with on-device reranking
- 95.2% R@5 recall on LongMemEval-S benchmark
- Sub-20ms P50 latency on a laptop
- 92% fewer input tokens per session
- 54 MCP tools (memory_save, memory_recall, memory_smart_search, etc.)
- 130 REST endpoints under /agentmemory/*
- Hourly consolidation sweeps: compress, merge duplicates, decay stale rows, emit audit rows
- Knowledge graph extraction with temporal edges
- Mesh federation: peer-to-peer sync over authenticated HTTPS
- Obsidian export with frontmatter-tagged markdown
- JSONL session import (Claude Code transcripts) for replay and crystal extraction
- OTEL observability with traces and logs (Jaeger/Honeycomb/Tempo export)
- Runs as single Node.js process, zero external databases (no Redis, Kafka, Postgres)
- Real-time viewer on port 3113 and engine console on port 3114
About Agentmemory
Agentmemory is an open-source (Apache-2.0) memory runtime that gives AI coding agents a persistent memory that survives the session. Built for developers using Claude Code, Cursor, Copilot CLI, and other MCP-compatible agents, it captures every prompt, tool call, and stop through 12 auto-hooks, compresses them into observations, and recalls relevant memories with a hybrid retriever that ranks BM25, vector, and knowledge-graph scores together and reranks them on-device. The result: 95.2% R@5 recall on LongMemEval-S with about 92% fewer input tokens per session than full-context replay. Install with one command (`npx @agentmemory/agentmemory`), start the server, and connect your agent with `agentmemory connect <agent>`. Native plugins exist for Claude Code, Copilot CLI, Codex CLI, OpenClaw, Hermes, pi, and Cursor; every other MCP client gets the full MCP server for free. The runtime exposes 54 MCP tools and 130 REST endpoints under /agentmemory/*, so you can query memory from curl, a browser, or your own agent. Consolidation runs automatically with an LLM provider key: raw observations compress into semantic memories, duplicates merge, stale rows decay, and audit rows record it all. A real-time viewer on port 3113 shows live observation streams, session replay, memory browsing, and knowledge graph visualization; an engine console on port 3114 displays OTEL spans and logs. You can federate multiple nodes over authenticated HTTPS, export memories to Obsidian as frontmatter-tagged markdown, and import Claude Code JSONL transcripts for replay. Bring your own Claude subscription, or point it at Anthropic API, Gemini, MiniMax, or OpenRouter—Ollama works for offline use. The v0.9.29 release adds a Cursor plugin with 7 auto-capture hooks and MCP version negotiation. Verified by 1,718 passing tests and backed by the Linux Foundation, Agentmemory is a complete runtime rather than a library or vector store, with zero external dependencies and built-in observability.
Behind the Verdict
Agentmemory is a serious, well-engineered answer to the context-loss problem plaguing AI coding agents. It's a full runtime, not a vector store or a library—you get capture, recall, consolidation, observability, and federation in one Node.js process with zero external services. Strengths: The 12 auto-capture hooks are the most complete we've seen, catching every prompt, tool call, and stop with no glue code. The hybrid recall path (BM25 + vector + knowledge graph) with on-device reranking is genuinely different from single-index solutions, and the 95.2% R@5 on LongMemEval-S is benchmark-proven and reproducible. The 54 MCP tools and 130 REST endpoints give you an astonishingly complete surface, and the five origin channels (user, agent, tool, import, shared) provide real provenance. Weaknesses: It's not a managed cloud service, so you're responsible for running, backing up, and securing the server. There's no built-in RBAC; access control is primarily via bearer tokens for federation. The CLI and MCP configuration may intimidate non-developers, but if you're using Claude Code or Cursor, you're already comfortable with that. Where it fits: Developers and teams working on large codebases with long-running autonomous agents, especially in self-hosted or offline environments. The Obsidian export and JSONL import make it easy to integrate with your knowledge management workflow. Where it doesn't: If you need a zero-ops cloud solution or enterprise-grade permissions, look at mem0 or Letta. But for local-first, open-source control, Agentmemory is the strongest option we've tested.
Researching Agentmemory? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Agentmemory actually fits — and what changes day-one when you adopt it.
You're working on a large codebase and often lose context between sessions. You install Agentmemory via npx and run 'agentmemory connect claude'.
Outcome: Within minutes, Claude Code automatically captures every prompt and tool call, and you can recall past decisions with memory_recall, saving thousands of tokens per session.
Your team uses Cursor and wants a shared memory of architectural decisions. You register a peer node and federate memories over HTTPS.
Outcome: Each developer's agent contributes to a shared memory pool, and consolidation merges duplicates, ensuring decisions are preserved and retrievable across the team.
Use Cases
- Install agentmemory via npm and connect Claude Code in under 2 minutes for persistent context across sessions.
- Use the triple-stream recall to instantly retrieve relevant code decisions and commands from past agent sessions.
- Automatically capture every tool call and prompt as observations, then consolidate them into semantic memories hourly.
- Export compressed memories as Obsidian-compatible markdown for human review and knowledge management.
- Federate memories across team members by registering a peer node and syncing over authenticated HTTPS.
- Import a Claude Code JSONL transcript to rehydrate a full session including observations, tool uses, and timeline.
Models Under the Hood
as of 2026-09-01
Limitations
- Agentmemory is an open-source, local-first persistent memory runtime for AI coding agents.
- It captures what your coding agent does, distills it into durable memory, and recalls it in later sessions, all running on your machine with zero external databases.
- The runtime is built on the iii engine and runs as a single process, offering 54 MCP tools, 130 REST endpoints, and 12 auto-hooks.
- Capture and recall work without an LLM key; compression, summarization, consolidation, and graph extraction activate when a provider key is set.
as of 2026-08-31
Verification history
We have re-verified Agentmemory 8 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 8 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Agentmemory tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source (Community)
$0/mo
Ideal for
Developers and small teams who want a free, self-hosted memory runtime for AI coding agents, are comfortable with CLI tools, and can provide their own LLM API key for advanced features.
What this tier adds
This is the only tier, available at $0/mo, offering all core features including 12 auto-hooks, hybrid recall, 54 MCP tools, 130 REST endpoints, and native plugins for various agents.
Where the pricing makes sense
The company stage and team size where Agentmemory's pricing actually pencils out — and where peers do it cheaper.
Agentmemory is completely free and open-source (Apache-2.0) with no paid tiers, making it ideal for individual developers and small teams who want a self-hosted memory runtime without subscription costs. Compared to managed options like mem0 or Letta, which typically charge per seat or usage, Agentmemory offers a zero-cost entry point, though you'll need to provide your own LLM API key for advanced features. This pricing fits developers who are comfortable running their own infrastructure and
Setup time & first value
How long it actually takes to get something useful out of Agentmemory — broken out by persona, not the marketing-page minute.
For a developer familiar with CLI tools: Installation with npx takes about 1 minute, and connecting an agent (e.g., Claude Code) is another minute. The server auto-starts the viewer on port 3113, so you can see observations immediately. Total time to first value: under 5 minutes. For teams setting up federation, allow an additional 10-15 minutes to configure HTTPS and bearer tokens.
Switching to or from Agentmemory
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Claude Code JSONL transcripts: Use the JSONL import feature to rehydrate past sessions and index them for search.
- →From manual note-taking: Export existing notes as markdown and import or sync them via Obsidian export for a unified view.
- ↗To mem0: If you need a managed cloud solution, you can export your memories via REST endpoints and manually map them to mem0's data structures.
- ↗To Letta: For a server-based memory with built-in agents, export memories as markdown or JSONL and import them into Letta's memory blocks.
Integrations
Resources & Guides
- Quickstartagent-memory.dev
Quickstart · Agentmemory
Get up and running fast from agent-memory.dev
- Documentationagent-memory.dev
Connectors · Agentmemory
Full product docs from agent-memory.dev
- Documentationagent-memory.dev
Integrations · Agentmemory
Full product docs from agent-memory.dev
- Documentationagent-memory.dev
How It Works · Agentmemory
Full product docs from agent-memory.dev
- Documentationagent-memory.dev
Configuration · Agentmemory
Full product docs from agent-memory.dev
- Documentationagent-memory.dev
Mcp Tools · Agentmemory
Full product docs from agent-memory.dev
- Documentationagent-memory.dev
Viewer · Agentmemory
Full product docs from agent-memory.dev
Tutorials & Learning
Official links
Tools that pair well with Agentmemory
Common stack mates teams adopt alongside Agentmemory, with the specific reason each pairing earns its keep.
Cognee
Open-source graph memory platform that gives AI agents persistent, relationship-aware recall
Distill
Open-source context intelligence & persistent memory layer for LLM agents with ~12ms deterministic dedup.
Powermem
Open-source, self-hosted memory layer that gives AI agents persistent, intelligent recall via hybrid retrieval.
Featured Head-to-Head Comparisons
Agentmemory vs Spider Cloud
Choose Agentmemory if you need persistent memory for coding agents across sessions, especially with Claude Code or MCP tools. Choose Spider Cloud if your AI agents or RAG pipelines require real-time web data at low cost. Both are free-tier accessible but solve opposite problems: memory vs. retrieval.
Agentmemory vs Voyage Ai
Agentmemory is the no-brainer choice for developers needing persistent agent memory — it's free, self-hosted, and integrates directly with coding agents like Claude Code. Voyage AI wins if you need high-accuracy embeddings for domain-specific RAG (finance, legal), but its enterprise pricing and lack of transparency make it unsuitable for smaller teams. Choose by need: agent memory vs. search retrieval.
Agentmemory vs Temporal Ai
Agentmemory is the go-to for developers who need persistent memory across coding agent sessions with zero external dependencies and high recall. Temporal AI is better for teams building durable, long-running workflows that require automatic retries, human-in-the-loop, and fault tolerance. Choose based on whether your bottleneck is agent context loss (Agentmemory) or process reliability (Temporal).
Alternatives to Agentmemory
View allFrequently Asked Questions
Used Agentmemory? Help shape our editorial sentiment research.