Ryumem - Semantic Memory for AI Agents

Strategies

Semantic Search: Embedding similarity using vector embeddings.
BM25 Keyword Search: Traditional keyword/lexical matching.
Graph Traversal: Navigate relationships with BFS.
Hybrid: Combines all three using RRF fusion (recommended).

Usage

from ryumem import Ryumem

ryumem = Ryumem(server_url="http://localhost:8000")

# Try different strategies
results = ryumem.search(
    query="AI researchers",
    user_id="user_123",
    session_id="session_abc",
    strategy="semantic"
)

results = ryumem.search(
    query="machine learning",
    user_id="user_123",
    session_id="session_abc",
    strategy="bm25"
)

results = ryumem.search(
    query="tech companies",
    user_id="user_123",
    session_id="session_abc",
    strategy="hybrid"
)

Search Parameters

Parameter	Type	Required	Description
`query`	str	Yes	The search query
`user_id`	str	Yes	User identifier for multi-tenancy
`session_id`	str	Yes	Session identifier
`limit`	int	No	Maximum results to return (default: 10)
`strategy`	str	No	Search strategy: `semantic`, `bm25`, `traversal`, or `hybrid`
`similarity_threshold`	float	No	Minimum similarity score for results
`max_depth`	int	No	Graph traversal depth (default: 2)
`kinds`	list	No	Filter by episode kinds: `query` or `memory`