Streaming is one the most beautiful things in Scoopika, at least in my opinion :)

Instead of waiting for the LLM to finish generating the response, or for the agent to finish executing tools, we stream the events in real-time.

Scoopika provides +10 hooks to capture specific events during the agent process, like generated tokens, called tools, called tools results, audio chunks, and more.

This takes the agent response latency from few seconds to few milliseconds.

Streaming hooks work both on the server-side and client-side out of the box. example usage:

{
    onToken: (token) => {...},
    onToolCall: (call) => {...},
    onToolResult: (res) => {...},
    onAudio: (audio) => {...}
}

Was this page helpful?