# Feather DB

> Feather DB is a lightweight, embedded vector database + living context engine. Zero-server, file-based, powered by a C++ core with Python bindings and a Rust CLI. Current version: v0.8.0.

Feather DB is designed for AI engineers and LLM application developers who need fast, serverless vector storage with native multimodal support. It stores data in a single optimized `.feather` binary file — no Docker, no network calls, no infra overhead.

Key features introduced in v0.8.0:
- **Multimodal Pockets** — A single entity ID can hold `text`, `visual`, and `audio` vectors simultaneously, each indexed in its own scoped HNSW index.
- **Context Graph & Chain** — Store vector relationships natively. `db.context_chain()` combines semantic ANN search with N-hop BFS graph traversal in a single call.
- **Adaptive Decay** — Vectors decay in relevance over time based on a configurable half-life. Older context naturally yields to newer, more relevant data.

## Docs

- [Overview](https://getfeather.store/docs): What is Feather DB, core concepts, and version notes.
- [Installation](https://getfeather.store/docs/installation): Install via `pip install feather-db` (Python) or `cargo install feather-db-cli` (Rust CLI).
- [Quick Start](https://getfeather.store/docs/quickstart): Create a DB, add multimodal vectors, search, and persist in under 5 minutes.
- [API Reference](https://getfeather.store/docs/api): Full reference for `DB.open()`, `db.add()`, `db.search()`, `db.context_chain()`, and more.
- [Storage Format](https://getfeather.store/docs/storage): Binary `.feather` file format layout, magic numbers, version blocks, and HNSW edge encoding.
- [Integrations](https://getfeather.store/docs/integrations): How to use Feather DB with LangChain, LlamaIndex, and custom embedding pipelines.

## Theory (Blog)

- [The Context Layer Performance Marketing Actually Needs](https://getfeather.store/theory/the-context-layer-performance-marketing-actually-needs): Why context is now the missing link in AI-driven marketing systems.
- [How Agents Can Use Filesystems for Context Engineering](https://getfeather.store/theory/how-agents-can-use-filesystems-for-context-engineering): Filesystem-native approaches to LLM memory and context management.
- [Introducing Feather Cloud](https://getfeather.store/theory/introducing-feather-cloud): Announcement and architecture overview of the Feather Cloud hosted offering.

## Optional

- [GitHub Repository](https://github.com/feather-store/feather): Source code, issues, and contribution guide.
- [PyPI Package](https://pypi.org/project/feather-db/): Python package page with release history.
- [Crates.io](https://crates.io/crates/feather-db-cli): Rust CLI crate page.