Backends and databases
Backends, databases and storage SDKs for apps and agents. 6 resources; API, key, license and embedding quoted from each site's own pages, checked Oct 24, 2026.
Files SDK
You can switch your app from S3 to R2 or Vercel Blob by changing one adapter line instead of rewriting every upload and signed-URL call.
Open sourceFree tierCommercial OK
Powabase
You get database, auth, file storage, RAG search and tool-calling agents from one backend instead of wiring Postgres, a vector database, an agent framework and glue code together yourself.
APIOpen sourceFree tierCommercial OK
Fluree
You can give an AI agent a knowledge graph with typed relationships, record history and per-user access rules, so answers can be traced to specific facts instead of loose vector matches.
APIFree tierCommercial OK
SurrealDB
An app or agent that needs vector search, graph traversal and plain records at once can keep them in one database and one transaction instead of wiring three separate stores together.
APICommercial OK
Directus
You get a full API and an editing UI for an existing Postgres or MySQL database without writing a CRUD backend or admin panel.
APIDownloadFree tierCommercial: conditions
Busabase
What your agents produce stops vanishing into chat logs: it lands in structured records the next agent or teammate can read, and writes from untrusted keys can wait for a person to approve.
APIOpen sourceFree tierCommercial OK
Side by side
| Resource | API | Key | Data download | Commercial use | Embed |
|---|---|---|---|---|---|
| Files SDK An MIT-licensed TypeScript library (npm i files-sdk) that gives one set of methods (upload, download, list, search, signed URLs and more) over 48 storage backends such as S3, R2, GCS, Azure Blob, Vercel Blob and Supabase, with plugins for encryption or versioning, React/Vue/Svelte upload UI, a CLI and a read-only MCP server. |
No official API | — | — | Commercial use OK | Embedding unclear |
| Powabase A Supabase-style backend for AI apps: each project gets its own Postgres with pgvector, auth, storage and realtime, plus a document pipeline that OCRs, chunks, embeds and indexes uploads, an agent runtime with HTTP/MCP tools and visual workflows, all behind one REST API; hosted or self-hosted from an Apache-2.0 repo. |
Official API · Free tier | Yes | — | Commercial use OK | Embedding unclear |
| Fluree A graph database (FlureeDB) that stores data as JSON-LD/RDF with history and access policies, queried with SPARQL or FlureeQL over HTTP, plus a hosted platform (Fluree AI) that adds connectors, an MCP server and natural-language answers over the graph. |
Official API · Free tier | Not stated | — | Commercial use OK | Can be embedded |
| SurrealDB A database engine written in Rust that stores documents, graph relations, vectors, time-series and relational tables in one place and queries them all with one SQL-like language (SurrealQL); it runs embedded, on one server, as a cluster, or as a hosted cloud service. |
Official API · Free tier | Yes | — | Commercial use OK | Embedding unclear |
| Directus A backend you run (self-hosted with Docker or on Directus Cloud) on top of any SQL database: it generates REST and GraphQL APIs, an admin web app for editing data, auth and roles, file storage, automations, realtime WebSockets and a built-in MCP server. |
Official API · Free tier | Not stated | Yes | Commercial use with conditions | Embedding unclear |
| Busabase An open-source (MIT) shared workspace where AI agents and people read and write typed tables, docs, skills and small apps over MCP, a REST/OpenAPI API, a CLI or SDKs, with each agent write recorded as a change request with author, diff and history. |
Official API · Free tier | Yes | — | Commercial use OK | Embedding unclear |