Cache¶
EnrichMCP includes a simple caching system available from EnrichContext.cache.
It stores values in namespaces scoped to the current request, authenticated user, or application.
The cache uses pluggable backends so you can keep data in memory or an external store like Redis.
ContextCache¶
The ContextCache class is created for every request. Use its helper methods to
store or retrieve data:
value = await ctx.cache.get_or_set("key", factory, scope="request", ttl=None)
await ctx.cache.set("key", value, scope="user")
cached = await ctx.cache.get("key", scope="global")
await ctx.cache.delete("key")
Cache namespaces are generated automatically using your app's ID, the request ID and, for user scope, a hash of the access token:
enrichmcp:global:{app_id}
enrichmcp:user:{app_id}:{user_hash}
enrichmcp:request:{app_id}:{request_id}
Default TTLs are global=3600, user=1800, and request=300 seconds. If the
user scope is requested but no access token is available a warning is emitted
and the key is stored in the request scope instead.
Backends¶
Two backends are provided:
MemoryCache– in-memory storage used by defaultRedisCache– persistent storage backed by Redis
::: enrichmcp.cache.ContextCache options: show_source: true
::: enrichmcp.cache.MemoryCache
::: enrichmcp.cache.RedisCache