GPT Commands

GPT commands make requests to an LLM endpoint. In our grammar, each GPT command is prefixed with the word model. A model command has 4 parts

model
A prompt name
A source (optional; defaults to selected text)
A destination (optional; defaults to paste)

You can mix and match any combination of prompt, source, and destination.

Examples

Command	Explanation
`model fix grammar`	Fix typos in the selected text and paste the response. `fix grammar` is the prompt name.
`model fix grammar to clip`	Fix typos in the selected text and put the result on the clipboard. `to clip` is a destination.
`model explain clip to window`	Explains the text on the clipboard and puts the response in a new browser window. `clip` is the source. `to window` is the destination.

Other model commands

There are also a few other special commands like model please and model ask that don’t use a prompt, but instead use the user’s arbitrary natural language request.

Command	Description	Example
`model help`	Show the help menu with all the prompts	”model help”
`model please <text>`	Say an arbitrary prompt and then apply it	”model please translate this to Japanese”
`model ask <text>`	Ask a question to the model	”model ask what is the meaning of life”

All model options

The left is the spoken word, and the right is how talon interprets it

Mappings of prompts to their spoken name

list: user.staticPrompt
-

# Use static prompts as aliases for detailed instructions
# Reduce verbosity and prevent the need to say the entire prompt each time time

## FIXES
fix grammar formally: Fix any mistakes or irregularities in grammar, spelling, or formatting. Use a professional business tone. The text was created using voice dictation. Thus, there are likely to be issues regarding homophones and other misrecognitions. Do not change the original structure of the text.
fix grammar: Fix any mistakes or irregularities in grammar, spelling, or formatting. The text was created using voice dictation. Thus, there are likely to be issues regarding homophones and other misrecognitions. Do not change the tone. Do not change the original structure of the text.
fix syntax: Fix any syntax errors in this code selection. Do not change any behavior.

## FORMATTING
format table: The following markdown text is raw data. There is no index. Return the text in a markdown table format. Each row has a new line in the original data.
format bullets: Convert each paragraph into a heading with a series of bullet points underneath it. Each paragraph is separated by a new line. Separate paragraphs should not have combined bullet points. This should all be done in markdown syntax. If it is a small paragraph, then you can just leave it as a heading and not add bullet points. Do not reduce content, only reduce things that would be redundant. These bullet points should be in a useful format for notes for those who want to quickly look at it. If there is a citation in the markdown original, then keep the citation just at the top and not within every individual bullet point.
format mermaid: Convert the following plain text into the text syntax for a mermaid diagram.
format comment: Format the following text as a comment for the current programming language. Use the proper comment syntax for the current language. Split the comment into multiple lines if the lines are too long.
group: Act as an organizer. The following text consists of various topics all put together. Please group these items into categories and label each category. Return just the results.
join: Act as an editor. The following text is separated into multiple parts. Please group them together into one part maintaining the flow and meaning. Reorder in whatever way makes sense. Remove any redundant information. The result should be only one part with no additional structure. Return just the modified text.

## TEXT GENERATION
explain: Explain this text in a way that is easier to understand for a layman without technical knowledge.
summarize: Summarize this text into a format suitable for project notes.
add context: Add additional text to the selected text that would be appropriate to the situation and add useful information.
fit schema: The given text has a series of responses that need to be categorized. Each response has a key that needs to be mapped to a value. Infer the schema from the text unless it is given at the top of the text with prior examples. Return the key-value pairs in a JSON format unless you infer a different format.
answer: Generate text that satisfies the question or request given in the input.
shell: Generate a shell script that performs the following actions. Output only the command. Do not output any comments or explanations. Default to the bash shell unless otherwise specified.
add emoji: Return the same exact text verbatim with the same formatting, but add emoji when appropriate in order to make the text fun and easier to understand.
make softer: Act as an editor. I want you to make the following text softer in tone. Return just the modified text.
make stronger: Act as an editor. I want you to make the following text stronger in tone. Return just the modified text.

## FILE CONVERSIONS
convert to jason: Convert the following data into a JSON format.
convert to markdown: Convert the following text into a markdown format.
convert to python: Convert the following key-value pairs into the syntax for a Python dictionary. So you should serialize the key-value pairs into a native Python format.
convert to sheet: Convert the following data into a CSV format.
convert to yam: Convert the following data into a YAML format.

## CHECKERS
describe code: Explain what the following code does in natural language at a high level without getting into the specifics of the syntax.
check grammar: Check the grammar and formatting of the following text. Return a list of all potential errors.
check spelling: Check the spelling of the following text. Return a list of all potential errors.
check structure: Skim the structure and layout of the following text. Tell me if the structure and order of my writing are correct. If it is not correct or flows poorly, then tell me what might be wrong with it. If it is all correct, then say it looks good.

## TRANSLATIONS
translate to english: Translate the following text into English.

## CODE GENERATION
generate code: The following plaintext describes a process in code in the language that is specified by the system prompt. Please output the code necessary to do this. Return just code and not any natural language explanations.
update comments: Act as a software engineer. The following code may be missing comments or the comments could be out of date. Please update the comments. If you are unsure how to comment something, ask a question in a comment instead. Return just the code and not any explanations.
clean code: Act as a software engineer. Reduce any duplication in the selected code and improve it to be more idiomatic and clear for other users. However, do not change the behavior or functionality. Return just the code and not any explanations.
improve semantics: The following is an HTML document. Keep the same structure and layout but if it is needed, change any elements to use proper semantic HTML and make sure it is implementing best practices for user accessibility. Output just the HTML and not any extra explanations.

## WRITING HELPERS
add questions: Help me explore this question from multiple perspectives. For each perspective, ask follow-up questions and indicate what perspective is being taken.
format outline: Create an outline that encapsulates the text below. Keep the number of sections between three and five to optimize for human working memory. Return just the outline.
format prose: As an editor, format the following outline or summarization as prose. You can have headings and paragraphs. Avoid using bullet points. Reorder and add transitions as necessary to make the document flow. Return just the text.

Sources that the model can operate on

list: user.modelSource
-

# A list of sources from which the model can take in data to process

# Apply the prompt to the text on the clipboard
clip: clipboard

# Apply the prompt to the text in the context
context: context

# Apply the prompt to the currently selected text
this: selection

# Apply the prompt to the previously returned GPT response
response: gptResponse

# Apply the prompt to the last phrase typed by Talon via a user's dictation
last: lastTalonDictation

Destinations for model insertions

list: user.modelDestination
-

# A list of adjectives/adverbs describing how the result of the GPT query should be returned

# paste in place
to this: paste

# Paste above the current selection
above: above

# Past below the current selection
below: below

# Instead of pasting, return the result to the clipboard
to clip: clipboard

# Instead of pasting, add the result to the context
to context: context

# Instead of pasting, add the result to a new context
to new context: newContext

# Instead of pasting, append the result to the clipboard
to after clip: appendClipboard

# Open the result in the browser
to browser: browser

# Speak the result with TTS
to speech: textToSpeech

# Output the result in a talon imgui window
to window: window

# Select the response after insertion so you can apply subsequent prompts on it
chain: chain

# Insert the response as a snippet with placeholders (only works in vscode)
snip: snip