Moss Samples

Moss is a high-performance runtime for real-time semantic search. It delivers sub-10 ms lookups, instant index updates, and zero infra overhead. Moss runs where your agent lives - cloud, in-browser, or on-device - so search feels native and users never wait. You connect your data once; Moss handles indexing, packaging, distribution and updates.

This repo bundles thin, working examples that show how to talk to Moss from Python and JavaScript. Each sample keeps the scaffolding light so you can copy the essentials straight into your own projects.

Join our discord server to get onboarded!

Using Moss Portal

• Visit usemoss.dev - click on signup/login on the top right
  • confirm your email, and sign in.
• Inside the default project you will see two plans:
  • Free Tier ($0) offers 1 project, 3 indexes, and 1,000 items per index with community support;
  • Developer Workspace ($2000/month + usage) adds unlimited projects/indexes plus 100 GB storage, 100 GB ingestion, 1 TB egress, and priority support.
• Enter valid card details to start the free trial, then select Create Index to provision a new index.
• From the dashboard, open View secrets and save the values as MOSS_PROJECT_ID and MOSS_PROJECT_KEY in your .env for the samples.

Building with Moss

• Samples show how to authenticate, batch context, and stream replies without extra boilerplate.
• Adapt the scripts by swapping the FAQ JSON files with your data, or plugging Moss calls into an existing app.

Python SDK Quick Tour

• python/comprehensive_sample.py: end-to-end flow with session creation, context building, and streaming responses.
• python/load_and_query_sample.py: how to ingest domain knowledge before querying Moss.
• python/custom_embedding_sample.py: create a fresh index, add custom OpenAI embeddings, and run sample queries.
• Install deps with pip install -r python/requirements.txt, then run any script via python path/to/sample.py.

JavaScript SDK Quick Tour

• javascript/comprehensive_sample.ts: TypeScript version of the full workflow, ready for Node.
• javascript/load_and_query_sample.ts: demonstrates indexing FAQs and issuing targeted prompts.
• javascript/custom_embedding_sample.ts: provision a new index, push OpenAI embeddings, and run sample queries.
• Install deps with npm install inside javascript/, then execute via npm run start -- path/to/sample.ts.

Next.js SDK Quick Tour

• next-js/: A modern, web-based semantic search interface using Next.js 15 and Server Actions.
• Demonstrates how to securely call Moss from the server while providing a sleek, responsive UI to the user.
• To run locally: cd next-js && npm install && npm run dev. Open http://localhost:3000.

Pipecat Voice Agent Quickstart

• pipecat-moss/pipecat-quickstart/: Voice bot that plugs Moss retrieval into Pipecat’s real-time pipeline.
• Shows how to ingest FAQs with create-index.py, wire pipecat-moss package into Pipecat, and deploy to Pipecat Cloud.
• Use it to bootstrap a sub-10 ms semantic search customer support agent. For deeper context, see the project README and the Pipecat-Moss repo.

Learn More

• API reference: MOSS docs
• Our Launch YC Post!

Contributing

We welcome contributions! Please see our Contributing Guide for more information.

If you spot gaps or want another language example, open an issue or PR. We track feedback closely.

License

This project is licensed under the BSD 2-Clause License.