Mem0

Memory layer for AI agents that extracts and recalls facts per user, agent or run.

Provider: default
Kind: Cache
Category: Data
Official docs: https://docs.mem0.ai/platform/platform-vs-oss
Also known as: Mem0 Platform, Mem0 Open Source

Properties · 3 set by this component

property	type	values	required	kind default	this component
deployment · component property	enum	platform \| open-source	no	platform	inherited
evictionPolicy	enum	lru \| lfu \| ttl \| none	no	—	inherited
graphMemory · component property	boolean	—	no	false	inherited
hashing · advanced	enum	none \| consistent	no	none	inherited
invalidation	enum	ttl \| event \| none	no	ttl	▸ event
managed	boolean	—	no	false	▸ true
materialised · advanced	boolean	—	no	false	inherited
persistence	boolean	—	no	false	▸ true
plan · component property	enum	hobby \| starter \| pro \| enterprise	no	hobby	inherited
readBarrier · advanced	boolean	—	no	false	inherited
replication	enum	none \| single-az \| multi-az \| multi-region	no	none	inherited
shards	integer	—	no	—	inherited
structure · advanced	enum	key-value \| bloom-filter \| hash-set \| sorted-set	no	key-value	inherited
tier · advanced	enum	core \| edge	no	core	inherited
ttlJitter · advanced	boolean	—	no	false	inherited
ttlSeconds	integer	—	no	—	inherited
usedAsPrimaryStore · advanced	boolean	—	no	false	inherited

Rules that apply · 3

Every rule whose selector covers this component's kind or its category. Whether a rule fires depends on the design: its properties, its edges and its groups.

Cache used as primary store without persistenceerror

cache-as-primary-store-without-persistence

This cache is the only copy of its data and has persistence turned off. A restart loses the data outright.

— all of these hold:
— the node's `usedAsPrimaryStore` is true
— the node's `persistence` is false

Cache without eviction policywarning

cache-missing-eviction-policy

A cache with no declared eviction policy inherits whatever the vendor chose, and most vendors default to evicting nothing. The cache fills up and then starts refusing writes.

— any of these hold:
— the node's `evictionPolicy` has no value
— the node's `evictionPolicy` is "none"

Materialised cache without a rebuild patherror

materialised-cache-without-rebuild-path

A cache holding materialised views has to be rebuildable from a durable store or a worker. With nothing feeding it, a cold cache is a permanent gap rather than a slow start.

— all of these hold:
— the node's `materialised` is true
— this does not hold:
— the node has at least one incoming edge from a sql-database, nosql-database, worker or stream

Use it in a design

nodes:
  - { id: "mem0", component: "mem0" }

Other Cache products · 36

ApsaraDB for Memcachealibaba-apsaradb-memcachealibabacloudcache ApsaraDB for Redisalibaba-apsaradb-redisalibabacloudcache Amazon DynamoDB Accelerator (DAX)aws-daxawscache Amazon ElastiCacheaws-elasticacheawscache Amazon MemoryDBaws-memorydbawscache Azure Cache for Redisazure-cache-for-redisazurecache Azure Managed Redisazure-managed-redisazurecache Generic cachecachecache Cache clustercache-clustercache cogneecogneecache Dragonflydragonflycache EVCacheevcachecache Microsoft Garnetgarnetcache VMware GemFiregemfirecache Graphitigraphiticache Hazelcasthazelcastcache Apache Igniteignitecache Infinispaninfinispancache KeyDBkeydbcache Lettalettacache Memcachedmemcachedcache Redisrediscache Redis Cloudredis-cloudcache Squidsquidcache Trickstertrickstercache Upstash Redisupstashcache Valkeyvalkeycache Varnish Cachevarnishcache Zepzepcache Memorystore for Redisgoogle-memorystoregcpcache Memorystore for Memcachedgoogle-memorystore-memcachedgcpcache Memorystore for Redis Clustergoogle-memorystore-redis-clustergcpcache Memorystore for Valkeygoogle-memorystore-valkeygcpcache IBM Cloud Databases for Redisibm-databases-for-redisibmcache OCI Cache with Redisoci-cache-redisocicache OCI Cache with Valkeyoci-cache-valkeyocicache

Source

1# yaml-language-server: $schema=../../schema/component.schema.json
2# Mem0: memory layer for AI agents that extracts facts from conversations and recalls them per user, agent or run.
3# Per "Platform vs Open Source", the hosted Platform needs no vector store, LLM or embedder and alone offers graph memory,
4# memory decay and webhooks; the open-source library and server bring their own stores.
5schemaVersion: "1"
6type: "component"
7id: "mem0"
8kind: "cache"
9provider: "default"
10docs: "https://docs.mem0.ai/platform/platform-vs-oss"
11aliases: ["Mem0 Platform", "Mem0 Open Source"]
12defaults:
13  managed: true
14  persistence: true
15  invalidation: "event"
16properties:
17  deployment:
18    type: "enum"
19    values: ["platform", "open-source"]
20    default: "platform"
21  plan:
22    type: "enum"
23    values: ["hobby", "starter", "pro", "enterprise"]
24    default: "hobby"
25  graphMemory:
26    type: "boolean"
27    default: false

The same page as markdown, for an agent: /catalogue/components/mem0.md