Quickstart

pip install cortexadb

from cortexadb import CortexaDB
from cortexadb.providers.openai import OpenAIEmbedder

# With an embedder (recommended) - auto-embeds text
db = CortexaDB.open("agent.mem", embedder=OpenAIEmbedder())

# Or with manual embeddings
db = CortexaDB.open("agent.mem", dimension=128)

# Auto-embedding (requires embedder)
mid1 = db.remember("The user prefers dark mode.")
mid2 = db.remember("User works at Stripe.")

# With metadata
mid3 = db.remember("User's name is Alice.", metadata={"source": "onboarding"})

# Semantic search
hits = db.ask("What does the user like?")
for hit in hits:
    print(f"ID: {hit.id}, Score: {hit.score:.3f}")

# Retrieve full memory
mem = db.get_memory(hits[0].id)
print(mem.content)  # b"The user prefers dark mode."

db.connect(mid1, mid2, "relates_to")
neighbors = db.get_neighbors(mid1)
# [Edge(to=mid2, relation="relates_to")]

# Chunk and store a document
db.load("document.pdf", strategy="recursive")

# Or ingest raw text
db.ingest("Long article text here...", strategy="markdown")

agent_a = db.namespace("agent_a")
agent_a.remember("Agent A's private memory")
hits = agent_a.ask("query only agent A's memories")

[dependencies]
cortexadb-core = { git = "https://github.com/anaslimem/CortexaDB.git" }

use cortexadb_core::CortexaDB;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let db = CortexaDB::open("/tmp/agent.mem", 128)?;

    // Store a memory with an embedding
    let embedding = vec![0.1; 128];
    let id = db.remember(embedding.clone(), None)?;

    // Query
    let hits = db.ask(embedding, 5, None)?;
    for hit in &hits {
        println!("ID: {}, Score: {:.3}", hit.id, hit.score);
    }

    // Connect memories
    let id2 = db.remember(vec![0.2; 128], None)?;
    db.connect(id, id2, "related_to")?;

    // Checkpoint for fast recovery
    db.checkpoint()?;

    Ok(())
}