oneiron

Embedded retrieval engine for memory-first applications.

One binary. One process. Zero network hops.

Why

Most retrieval stacks bolt together separate services for vectors, text, and graphs — network hops, consistency gaps, and operational complexity that doesn't belong on a phone. Oneiron runs in-process as a Rust library with C FFI bindings. Every query touches a single LMDB environment with ACID transactions. Embed it on iOS, Android, desktop, or a server.

Signals

Signal	Engine	What it finds
Vector	HNSW (flat NSW), SIMD-accelerated	Semantically similar content
Text	BM25 inverted index	Exact keywords and phrases
Graph	Personalized PageRank over typed edges	Relationally connected entities
Temporal	Bi-temporal range indexes	Events by when they happened or were recorded
Phonetic	Code-based posting lists	Fuzzy matches from voice/ASR misspellings

Any subset of signals can be combined via Reciprocal Rank Fusion with per-signal boosts.

Quick Start

use oneiron::{Vault, VaultConfig, EntityId};

let config = VaultConfig::device();
let vault = Vault::open("./my-vault", config)?;

let id = EntityId::now();
vault.put_entity(&id, b"msgpack blob")?;
vault.put_vector(&id, &embedding)?;

Install

Install the local daemon from crates.io after publication:

cargo install oneiron-server

From a checkout:

cargo install --path crates/oneiron-server

One-line install from GitHub:

curl -fsSL https://raw.githubusercontent.com/oneiron-dev/oneiron/main/deploy/install-oneiron-server.sh | sh

Create and inspect the default local vault:

oneiron-server init ~/.local/share/oneiron/default
oneiron-server doctor ~/.local/share/oneiron/default

Run the daemon:

oneiron-server serve --vault-path ~/.local/share/oneiron/default

serve also remains the default command for the existing flags:

oneiron-server --vault-path ~/.local/share/oneiron/default --port 9090

The default service vault convention is ~/.local/share/oneiron/default/. oneiron-server serve reads ~/.config/oneiron/oneiron.toml when present; file values are overridden by ONEIRON_* environment variables and then by CLI flags.

Building

cargo build --release
cargo nextest run -p oneiron                                # fast tier
cargo nextest run -p oneiron --features sync --profile full # full tier (CI parity)

Tests run via cargo-nextest (cargo install cargo-nextest); profiles and the slow-test tier live in .config/nextest.toml. Plain cargo test still works.

Upgrade Notes

• ANALYZER_VERSION = "v2" changes analyzer-manifest hashes to capture Han whichlang routing behavior. Existing text indexes built with older analyzer manifests must be rebuilt after upgrading; create a VaultConfig, set config.skip_text_index_manifest_check = true, reopen with that config, run MaintenanceBuilder::clear_text_index, reopen normally, then reindex documents.

Design

18 LMDB databases per vault. Atomic multi-database writes via BatchBuilder. MessagePack entity blobs. Context packing into LLM-ready formats.

Full details in the design docs:

• SCHEMA-DESIGN.md — database layout, key formats, encoding
• BUILD-PROMPT.md — architecture, algorithms, API surface
• DEPLOYMENT.md — local daemon install, config, and service templates

License

Apache 2.0