Quick Start

Quick Start

Prereqs

• Rust stable matching rust-toolchain.toml / workspace rust-version 1.88.
• Git.
• For real LLM-backed operation: OpenAI/Azure/OpenAI-compatible credentials.
• Optional: Jina API key for cross-encoder reranking.

Clone

git clone https://github.com/5queezer/karta.git
cd karta

Build and test

cargo fmt --all -- --check
cargo clippy --workspace --all-targets -- -D warnings
cargo test --workspace --lib
cargo test -p karta-core --test synthetic_memory_eval

Library usage

use karta_core::{Karta, config::KartaConfig};

#[tokio::main]
async fn main() {
    let config = KartaConfig::default();
    let karta = Karta::with_defaults(config).await.unwrap();

    karta.add_note("Sarah prefers Slack notifications over email").await.unwrap();
    karta.add_note("Brightline requires all integrations on the approved vendor list").await.unwrap();

    let result = karta
        .ask("What should I know about Sarah's notification preferences?", 5)
        .await
        .unwrap();

    println!("{}", result.answer);

    let dream_run = karta.run_dreaming("workspace", "default").await.unwrap();
    println!("Dreams: {} attempted, {} written", dream_run.dreams_attempted, dream_run.dreams_written);
}

Minimal env

Standard OpenAI path:

export OPENAI_API_KEY=...
# optional
export OPENAI_API_BASE=https://api.openai.com/v1
export KARTA_EMBEDDING_MODEL=text-embedding-3-small
export JINA_API_KEY=...

Azure OpenAI path:

export AZURE_OPENAI_API_KEY=...
export AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
export AZURE_OPENAI_CHAT_MODEL=gpt-4o-mini
export AZURE_OPENAI_EMBEDDING_MODEL=text-embedding-3-small
export AZURE_OPENAI_API_VERSION=2025-04-01-preview

Run MCP server

See MCP Server for OAuth setup. Minimal command after env is set:

cargo run -p karta-server

MCP endpoint: /mcp.