streaming

Streaming

Demonstrates real-time token delivery using Adam's streaming callback. Instead of waiting for the full response, the on_stream callback fires for each token as the LLM generates it, allowing immediate display. The program asks for a short story and prints each token to stdout as it arrives, producing a typewriter-like effect.

Build & Run

# Create a .env file with your API key
echo "ANTHROPIC_API_KEY=sk-ant-..." > .env

make
./streaming

What It Demonstrates

• Setting a stream callback with adam_settings_set_stream
• Real-time token-by-token output via adam_stream_fn
• The is_done flag to detect the final chunk
• adam_run still returns the complete response even when streaming