From d5fbb6ef5957894eb5be0854bdb328a6774abddb Mon Sep 17 00:00:00 2001 From: Paul Buetow Date: Sun, 17 Aug 2025 09:05:45 +0300 Subject: cli: stream responses in hexai when supported (OpenAI, Ollama) - Add llm.Streamer optional interface - Implement ChatStream for OpenAI (SSE) and Ollama (JSON stream) - CLI uses streaming; LSP unchanged (non-streaming) - README: document streaming behavior for CLI --- README.md | 1 + 1 file changed, 1 insertion(+) (limited to 'README.md') diff --git a/README.md b/README.md index 7de7365..b17983d 100644 --- a/README.md +++ b/README.md @@ -58,6 +58,7 @@ Notes for `hexai` (CLI): - Prints LLM output to stdout. - Prints provider/model immediately to stderr, and a summary to stderr at the end (time, input bytes, output bytes, provider/model). - Default response style: short answers. If the prompt asks for commands, outputs only the commands with no explanation. Include the word `explain` anywhere in the prompt to request a verbose explanation. +- Streams output: when supported by the provider (OpenAI, Ollama), `hexai` streams tokens and prints them to stdout as they arrive. Copilot falls back to non-streaming. ### Hexai CLI behavior -- cgit v1.2.3