Overview
What is Agentset?
Agentset is a TypeScript SDK for the Agentset platform, an agentic RAG-as-a-service. It provides programmatic access to create and manage knowledge bases, ingest content, and perform semantic search.
How to use Agentset?
Install the npm package (npm install agentset, yarn add agentset, or pnpm add agentset), then import the Agentset class, initialize it with an API key, and use methods like namespaces.create(), ns.ingestion.create(), and ns.search().
Key features of Agentset
- Create and manage namespaces (knowledge bases)
- Ingest text content into a namespace
- Search across ingested content
- Typed error handling (
NotFoundError,UnauthorizedError) - Custom fetch implementation support
Use cases of Agentset
- Build a question‑answering system over proprietary documents
- Create a searchable knowledge base for customer support
- Enable agentic RAG pipelines with custom embedding models
- Manage multiple isolated knowledge bases within one project
FAQ from Agentset
How do I install the SDK?
Use npm, yarn, or pnpm: npm install agentset, yarn add agentset, or pnpm add agentset.
How do I initialize the client?
Import Agentset and create an instance with your API key: new Agentset({ apiKey: "your_api_key_here" }).
What runtime is required?
The SDK is a TypeScript package meant for Node.js environments (npm‑based projects).
How are errors handled?
The SDK throws typed errors such as NotFoundError and UnauthorizedError, which you can catch with instanceof checks.
Can I provide a custom fetch implementation?
Yes, pass a fetcher option (e.g., node-fetch) when initializing the Agentset client.
