The voice commands can be broadly categorized into two categories:

<a href="#top">Back to top</a>

- If the larger functionality is not generating the desired results, consider breaking down the problem into smaller pieces. For example, if we want to generate a rock-paper-scissors game, we can start by generating a series of functions that can be used to play the game. Then, we can use the generated functions to build the game.

- a function to randomly select between rock, paper, and scissors

- "lets put it all together" - oftentimes, this phrase is useful to write a function that can tie together the smaller functions we have generated earlier.

- If the goal is to synthesize a method body, consider using the following formulation: "a method `intended method name` that accepts parameters `parameter names with types` that `intended functionality` and optionally return `return type`"

- Since the extension relies on the GitHub Copilot infrastructure, it is important to note that the code completions responses are non-deterministic. Therefore, it is possible that the extension may return slightly different code completions for the same input. Thus, canceling the code completions and trying again may yield a different result.

- The underlying speech recognition technology estimates the end of an utterance based on the silence in the audio stream. Since different folks have different speaking styles, we have left the silence threshold as a configurable parameter. You can change the silence threshold by editing the `Hey, GitHub!` extension settings. The default value is 1.25 seconds. However, if you are feeling particularly adventurous, try setting the value to 0.5 seconds.