The serverless vector database
that lives in your AWS account.

from dynavec import Dynavec, DynavecConfig, Document
from dynavec.embeddings import OpenAIEmbedder

cfg = DynavecConfig(
    vector_bucket="my-vectors",
    index="docs",
    table="dynavec_docs",
    dimension=1536,
    region="us-east-1",
    auto_provision=True,        # creates bucket + index + table
)
db = Dynavec(cfg, embedder=OpenAIEmbedder(model="text-embedding-3-small"))

db.upsert([
    Document(id="a", text="Mitochondria power the cell.", metadata={"topic": "bio"}),
    Document(id="b", text="Rockets reach orbit at ~28,000 km/h.", metadata={"topic": "space"}),
], auto_metadata=True)

for hit in db.search("how do cells make energy?", top_k=3):
    print(hit.score, hit.id, hit.text)

# metadata pre-filter + client-side metric rescoring + MMR diversity
hits = db.search(
    "fast serverless storage on AWS",
    top_k=5,
    filter={"topic": "aws"},
    rescore={"cosine": 0.7, "manhattan": 0.3},   # weighted combination
    rerank="mmr",
)

# bring your own vector (no embedder required)
hits = db.search(vector=my_1536d_vector, top_k=10)

# one index, many tenants — clean isolation + even partitioning
kb = db.namespace("tenant-42")
kb.upsert([Document(id="1", text="private doc")])
kb.search("query scoped to this tenant only", top_k=4)

# stream results to an agent as pages arrive
for hit in db.search_stream("large query", top_k=100):
    handle(hit)

# attach meaning: entities, relations, and links to documents
db.graph_add_edge("acme", "competes_with", "globex", namespace="kb")
db.graph_link("acme", ["doc-1", "doc-2"], namespace="kb")

# GraphRAG: traverse structure first, then rank only related docs
hits = db.graph_search(
    "recent product launches",
    seed_entities=["acme"],
    hops=2,
    top_k=10,
)

from dynavec import Dynavec, SemanticCache, DynamoDBCache

# serve near-duplicate queries from memory (no vector search)
db = Dynavec(cfg, embedder=embedder, cache=SemanticCache(threshold=0.97))

# or a shared, durable cache in your own DynamoDB table (TTL expiry)
db = Dynavec(cfg, embedder=embedder, cache=DynamoDBCache(cfg))
# RedisCache(...) targets AWS ElastiCache for sub-millisecond shared cache

from dynavec.integrations.langchain import DynavecVectorStore

store = DynavecVectorStore(db, namespace="kb")
retriever = store.as_retriever(search_kwargs={"k": 4})

# framework-agnostic tool for LangGraph / CrewAI / Strands
from dynavec.integrations.tools import make_retriever_fn
retrieve = make_retriever_fn(db, top_k=4)   # fn(query: str) -> str

from dynavec.ingest import ingest, MCPResourceSource

# turn ANY MCP server (Notion, Confluence, Drive, ...) into a corpus:
# chunk, embed, and upsert its resources in one call
ingest(db, MCPResourceSource(mcp_session), namespace="kb")