Supermemory OpenSearch
https://github.com/supermemoryai/opensearch-ai- Category
- AI Tools
- Rank
- No. 1412Tools index
Previous survey · No. 1419 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- supermemoryai
- GitHub
- 1.3k stars
- Date
About
Open-source SearchGPT/Perplexity clone personalized to your saved knowledge and history.
What it does
This is a signed-in web application that sends queries to Brave Search and returns web results. It also records each query in a user-tagged Supermemory collection. Users can list, add, and delete stored memories, while the interface runs as an installable progressive web app.
Why it's ranked here
The project offers a clear, compact example of combining search, authentication, and managed memory in a Next.js application. The personalization claim is only partly demonstrated by the supplied code: searches are stored, but no shown retrieval step uses those memories to alter results or generate an answer.
What's good
The search path remains usable when saving a query to memory fails. Stored items are separated with email-based user tags, and users can create or remove memories. Search responses receive an hourly cache policy. The project also includes progressive web app registration and explicit Cloudflare Pages build, preview, and deployment commands.
Tradeoffs
Running it requires several external credentials, including Brave Search, Supermemory, Google authentication, and OpenAI keys. The supplied search action rejects unauthenticated users. More importantly, the shown code returns Brave results directly and never injects retrieved memories into ranking or synthesis. Memory deletion checks for a signed-in user but does not visibly verify that the requested item belongs to that user.
How to use it well
Use it as a starting point for engineers exploring authenticated search plus persistent user memory in a Next.js and Cloudflare stack. Treat the current implementation as a prototype to extend with memory-aware retrieval, synthesis, and stronger deletion authorization. It does not provide a reusable library, command-line tool, standalone search backend, or documented offline workflow.
Technical notes+
app/actions.ts constructs Supermemory with SUPERMEMORY_API_KEY; getSearchResultsFromMemory stores the query under email and opensearch container tags, then calls Brave Search with SEARCH_API_KEY and a one-hour Next.js revalidation policy. getMemories performs a list followed by one fetch per memory, while deleteMemory accepts an arbitrary memory ID after only checking that a user exists. app/page.tsx selects the edge runtime and obtains a session through auth. next.config.mjs wraps the application with next-pwa, enables Cloudflare development bindings, and disables ESLint failures during builds. cf-env.d.ts declares six required service and authentication secrets. package.json exposes Next.js development, build, start, Cloudflare Pages preview, deployment, and Wrangler type-generation scripts.
Observed
- Primary language
- TypeScript, with React TSX components and TypeScript configuration files.
- Packaging
- Private npm package containing a Next.js web application.
- Interface
- Authenticated web application with server actions; no CLI or reusable library interface is shown.
- Deployment surface
- Cloudflare Pages build, local preview, and deployment scripts are provided.
- Search provider
- Web queries are sent to the Brave Search API.
- Memory provider
- Supermemory stores, lists, retrieves, and deletes user-tagged memories.
- Application mode
- Progressive web app support is configured with automatic registration and skip-waiting behavior.
- License
- No license is stated in the provided repository text.
Read from README.md, package.json, lib/utils.ts, env.d.ts, cf-env.d.ts, next.config.mjs, postcss.config.js, tailwind.config.ts, app/Card.tsx, app/page.tsx, app/types.ts, app/Blobs.tsx, app/Globe.tsx, app/actions.ts, app/layout.tsx.
What it can do
Search through saved knowledge base
Natural language query → Relevant results from personal knowledge collection
Generate AI-powered answers from personal data
User question and saved content → Contextual response based on personal knowledge
Save and index web pages
URLs or web page content → Searchable indexed content in knowledge base
Store and organize personal documents
Documents, notes, or text files → Searchable personal knowledge repository
Track and search browsing history
Web browsing activity → Searchable history with context
Provide personalized search results
Search query and user's saved data → Ranked results prioritized by personal relevance
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.