Skip to main content

Core Setup

Get started with omnicache-ai using different backends and configurations. No framework-specific extras are needed for core functionality.

pip install omnicache-ai

1.1 In-memory cache (zero config)​

The simplest way to start -- an in-memory cache that requires no external services.

from omnicache_ai import CacheManager, InMemoryBackend, CacheKeyBuilder

manager = CacheManager(
    backend=InMemoryBackend(max_size=10_000),
    key_builder=CacheKeyBuilder(namespace="myapp"),
)

1.2 Disk cache (survives process restarts)​

Persist cached data to disk so it survives process restarts. Backed by SQLite via diskcache.

from omnicache_ai import CacheManager, DiskBackend, CacheKeyBuilder

manager = CacheManager(
    backend=DiskBackend(directory="/var/cache/myapp"),
    key_builder=CacheKeyBuilder(namespace="myapp"),
)
note

The DiskBackend accepts directory (not path) as its first argument.

1.3 From environment variables​

Configure the cache entirely through environment variables -- useful for containerised deployments.

export OMNICACHE_BACKEND=redis
export OMNICACHE_REDIS_URL=redis://localhost:6379/0
export OMNICACHE_DEFAULT_TTL=3600
export OMNICACHE_NAMESPACE=myapp
from omnicache_ai import CacheManager, OmnicacheSettings

manager = CacheManager.from_settings(OmnicacheSettings.from_env())

1.4 With full TTL policy per layer​

Set different time-to-live values for each cache layer so embeddings, retrieval results, session context, and LLM responses each expire at the right cadence.

from omnicache_ai import CacheManager, InMemoryBackend, CacheKeyBuilder, TTLPolicy

manager = CacheManager(
    backend=InMemoryBackend(),
    key_builder=CacheKeyBuilder(namespace="myapp"),
    ttl_policy=TTLPolicy(
        default_ttl=3600,
        per_type={
            "embed": 86_400,      # embeddings last 24 h
            "retrieval": 3_600,   # retrieval results last 1 h
            "context": 1_800,     # session context 30 min
            "response": 600,      # LLM responses 10 min
        },
    ),
)

1.5 With tag-based invalidation​

Attach tags to cached entries and invalidate groups of related entries in a single call.

from omnicache_ai import (
    CacheManager, InMemoryBackend, CacheKeyBuilder, InvalidationEngine
)

tag_store = InMemoryBackend()
manager = CacheManager(
    backend=InMemoryBackend(),
    key_builder=CacheKeyBuilder(namespace="myapp"),
    invalidation_engine=InvalidationEngine(tag_store),
)