# DreamDB > DreamDB is a storage and retrieval protocol for multimodal signals anchored to a shared timeline — append-only, content-addressed, and searchable by semantic features. ## Overview - [DreamDB](https://dreamdb.dreamlake.ai/index.md): A storage and retrieval protocol for multimodal signals anchored to a shared timeline. ## Getting Started - [Installation](https://dreamdb.dreamlake.ai/installation.md): Install the TypeScript SDK and connect to a DreamDB backend. - [Quick Start](https://dreamdb.dreamlake.ai/tutorial.md): End-to-end walkthrough: ingest, snapshot, query, time-travel, and sharded ingest. - [Browser Demo](https://dreamdb.dreamlake.ai/browser-demo.md): Explore datasets in the browser — semantic search, time-travel, and multi-camera playback. - [LLM-Readable Docs](https://dreamdb.dreamlake.ai/llm-readable.md): Every page is available as clean markdown, plus an llms.txt index, a full-corpus dump, and an importable agent skill. ## TypeScript SDK - [Overview](https://dreamdb.dreamlake.ai/typescript-sdk.md): @dreamlake/dreamdb — TypeScript client for the DreamDB protocol. - [Install](https://dreamdb.dreamlake.ai/typescript-sdk-install.md): Install @dreamlake/dreamdb in a Node or browser project. - [Quickstart](https://dreamdb.dreamlake.ai/typescript-sdk-quickstart.md): Your first DreamDB dataset in 30 lines of TypeScript. - [API Reference](https://dreamdb.dreamlake.ai/typescript-sdk-api.md): Every public export of @dreamlake/dreamdb with TypeScript signatures. - [Browser usage](https://dreamdb.dreamlake.ai/typescript-sdk-browser.md): Read DreamDB Spaces directly in the browser. No app server. - [React example](https://dreamdb.dreamlake.ai/typescript-sdk-react.md): Minimal React thumbnail grid backed by a live DreamDB Space. - [Semantic search](https://dreamdb.dreamlake.ai/typescript-sdk-semantic-search.md): Browser-side CLIP text encoder + Dataset.iterVector for natural-language image search. ## Specification - [Overview](https://dreamdb.dreamlake.ai/spec-overview.md): Vocabulary and conceptual model for the protocol. - [Data Model](https://dreamdb.dreamlake.ai/spec-data-model.md): The five entities and three Track kinds. - [Content Addressing](https://dreamdb.dreamlake.ai/spec-content-addressing.md): Address grammar and content-addressed storage paths. - [Time Encoding](https://dreamdb.dreamlake.ai/spec-time-encoding.md): High-precision timeline encoding. - [Spatial Indexing](https://dreamdb.dreamlake.ai/spec-spatial-indexing.md): LSH-cosine partitioning and vector bucket layout. - [Backend Interface](https://dreamdb.dreamlake.ai/spec-backend-interface.md): The HTTP contract any compliant object store must provide. - [Protocol Operations](https://dreamdb.dreamlake.ai/spec-protocol-operations.md): The eight verbs: ingest, snapshot, query, stream, branch, merge, gc, delete. - [Streaming Encapsulation](https://dreamdb.dreamlake.ai/spec-streaming-encapsulation.md): Object byte layouts and streaming-native encapsulation. - [Versioning](https://dreamdb.dreamlake.ai/spec-versioning-collab.md): Branching, merging, and collaborative workflows. - [Conformance](https://dreamdb.dreamlake.ai/spec-conformance.md): Test vectors and compliance criteria. - [Compaction](https://dreamdb.dreamlake.ai/spec-compaction.md): Garbage collection and storage reclamation. - [Fragment Packs](https://dreamdb.dreamlake.ai/spec-fragment-packs.md): Bundling small items into single S3 objects. - [Index](https://dreamdb.dreamlake.ai/spec-index.md): Navigation index and open questions for the DreamDB protocol specification. - [Vector Compression](https://dreamdb.dreamlake.ai/spec-vector-compression.md): Product quantization and learned codebook compression. - [Scalar Indexing](https://dreamdb.dreamlake.ai/spec-scalar-indexing.md): Native structured-metadata filters. - [Federation](https://dreamdb.dreamlake.ai/spec-federation.md): Cross-cluster queries and distributed coordination. - [Graph Indexing](https://dreamdb.dreamlake.ai/spec-graph-indexing.md): Vamana / DiskANN-family graph-based ANN. - [Streaming Extensions](https://dreamdb.dreamlake.ai/spec-streaming-extensions.md): Item chunking and adaptive bitrate delivery. - [Hybrid Retrieval](https://dreamdb.dreamlake.ai/spec-hybrid-retrieval.md): Query planning across vector and scalar indexes. - [Real-Time Freshness](https://dreamdb.dreamlake.ai/spec-streaming-freshness.md): Streaming updates and near-real-time ingestion. - [Schema Evolution](https://dreamdb.dreamlake.ai/spec-schema-evolution.md): Embedding migration and backward compatibility. - [Multi-Tenancy](https://dreamdb.dreamlake.ai/spec-multi-tenant.md): Tenant isolation, quotas, and access control. - [Encryption](https://dreamdb.dreamlake.ai/spec-encryption.md): Data-plane encryption at rest and in transit. - [Tombstones](https://dreamdb.dreamlake.ai/spec-tombstones.md): Record-level deletion and GDPR compliance. ## Design - [Dataset Platform](https://dreamdb.dreamlake.ai/design-dataset-platform.md): Versioned multimodal data lake architecture on DreamDB. - [Scope Boundaries](https://dreamdb.dreamlake.ai/design-scope-boundaries.md): What belongs in the protocol vs. the application layer. - [ML Training](https://dreamdb.dreamlake.ai/design-ml-training-tutorial.md): Training PyTorch models on DreamDB datasets. - [Sharded Ingest](https://dreamdb.dreamlake.ai/design-sharded-ingest.md): Parallel ingest via branch-and-merge. - [Scale Blockers](https://dreamdb.dreamlake.ai/design-10b-scale-blockers.md): Audit of blockers on the path to 10 billion records. - [Retrospective](https://dreamdb.dreamlake.ai/design-known-flaws-retrospective.md): Known flaws and lessons from the initial implementation. - [Fragment Packs](https://dreamdb.dreamlake.ai/design-fragment-packs.md): Design rationale for bundling small items per S3 object. - [Roadmap](https://dreamdb.dreamlake.ai/design-todo-roadmap.md): Implementation roadmap and milestone tracking. ## Reference - [Changelog](https://dreamdb.dreamlake.ai/changelog.md): Implementation changelog with spec references. - [Release Notes](https://dreamdb.dreamlake.ai/release-notes.md): Per-version release notes for the DreamDB protocol and SDK. ## 规范 (中文) - [概览](https://dreamdb.dreamlake.ai/spec-chn-overview.md): --- - [数据模型](https://dreamdb.dreamlake.ai/spec-chn-data-model.md): --- - [内容寻址与地址语法](https://dreamdb.dreamlake.ai/spec-chn-content-addressing.md): --- - [时间编码](https://dreamdb.dreamlake.ai/spec-chn-time-encoding.md): --- - [空间索引](https://dreamdb.dreamlake.ai/spec-chn-spatial-indexing.md): --- - [后端接口](https://dreamdb.dreamlake.ai/spec-chn-backend-interface.md): --- - [协议操作](https://dreamdb.dreamlake.ai/spec-chn-protocol-operations.md): --- - [流式封装](https://dreamdb.dreamlake.ai/spec-chn-streaming-encapsulation.md): --- - [版本化与协作](https://dreamdb.dreamlake.ai/spec-chn-versioning-collab.md): --- - [一致性](https://dreamdb.dreamlake.ai/spec-chn-conformance.md): --- - [Value](https://dreamdb.dreamlake.ai/spec-chn-value.md): 这是一个宏大的命题。如果我们将 DreamDB 视为一种“多模态数据的 Git + 全球寻址的 Serverless 数据库”，它的核心价值在于打破了数据存储、协作与 AI 检索之间的物理墙。