TinyCortex is a local-first AI memory engine, shipped as the open-source Rust crate tinycortex. It gives your agents a memory that works the way a brain does: it intelligently forgets noise so the model only reasons over what matters.

Every AI memory system you have used does the same thing — store everything, retrieve by similarity, hope for the best. The outcome is an agent that drowns in stale context: responses degrade and costs inflate. TinyCortex takes the opposite approach. Low-value memories decay over time, while the knowledge your users recall and interact with is reinforced and rises to the top. There is no manual cleanup and no context-window anxiety.

The engine ingests content, canonicalizes and chunks it, scores what is worth keeping, and compresses it into a hierarchical summary tree. Retrieval then serves a focused, explainable slice of long-term history — vector, keyword, graph, and tree search combined — instead of a noisy dump of everything ever stored.

{% hint style="info" %} This documentation covers the open-source Rust crate. The hosted TinyCortex platform (managed API, language SDKs) is a separate product in closed alpha — reach out for access. Crate-only vs. hosted-only capabilities are called out throughout. {% endhint %}

cargo add tinycortex

use tinycortex::memory::{InMemoryMemoryStore, MemoryInput, MemoryQuery, MemoryStore};

#[tokio::main]
async fn main() -> anyhow::Result<()> {
    let store = InMemoryMemoryStore::new();

    store
        .insert(MemoryInput::new("preferences", "User prefers dark mode"))
        .await?;

    let hits = store.search(MemoryQuery::text("theme preference")).await?;
    for hit in hits {
        println!("{:.3}  {}", hit.score, hit.record.content);
    }
    Ok(())
}

See Getting Started for the full walkthrough, or jump to the Architecture Overview to understand how the engine fits together.

• Intelligent noise filtering — memories that are not accessed decay; frequently recalled knowledge becomes durable. The store stays lean on its own.
• Interaction-aware — views, replies, reactions, and authored content all signal what matters.
• Local-first & inspectable — markdown files are the source of truth; SQLite, vectors, summary trees, and a git ledger are rebuildable derived indexes.
• Explainable retrieval — every hit carries a score breakdown across graph, vector, keyword, and freshness signals.
• Provenance & safety — every item carries source identity and a security taint (internal vs. external-sync).

Discord • Reddit • X • crates.io