OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Toucandb

ToucanDB is an embedded, local-first Python vector database that persists embeddings in SQLite and uses FAISS for similarity search.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Toucandb — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use ToucanDB when an application needs local vector persistence and retrieval, especially for semantic search or a RAG workflow, and can operate within its single-process database ownership model.

Who it is for

  • Python developers building local semantic search or RAG features.
  • Application teams that need vector storage and retrieval without operating a separate database server.
  • Developers who want to supply vectors directly or connect an available embedding adapter.

Documented uses

  • Persist, index, and retrieve high-dimensional vectors.
  • Search vectors by similarity, including searches filtered by metadata.
  • Index documents and retrieve source-attributed context for an application-provided RAG generator.
  • Use embedding retrieval as application memory.
Before you choose

Requirements & limits

The repository identifies the project as open source under the MIT license. Python 3.10 or newer is required.

  • SQLite provides durable persistence while FAISS supplies the search index; the README describes FAISS state as rebuildable.
  • The database is embedded and one process owns a database directory at a time.
  • The RAG pipeline is in the core package, while Sentence Transformers and OpenAI adapters use optional extras.
  • The core RAG pipeline accepts an injected generator rather than requiring a particular LLM framework.
  • Flat search is exact; HNSW and IVF are approximate and have distinct memory, tuning, and workload trade-offs.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Store and search vectors directly

Create a collection, upsert supplied vectors, and search it with a metadata filter.

Before you start

  • Python 3.10 or newer.
  • Install the core toucandb package.
  • Provide vectors with consistent dimensions and a collection schema.

Steps

  1. Open a ToucanDB database at a local path.

  2. Create a collection schema with its dimensions and selected index type.

  3. Upsert vectors with IDs and optional metadata.

  4. Search with a SearchQuery containing a query vector, result count, and optional metadata filter.

Expected outcome

The application receives matching vector records from its collection.

Guide limitations

  • ToucanDB is embedded, not a distributed service, and a database directory has one owning process at a time.
  • Approximate index types trade exactness or memory and tuning requirements differently from flat search.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
ToucanDB/ToucanDB · Official homepage title · retained in its source language
How we review sourcesSuggest a correction