AgentFlow is a prompting language designed to generate LLM (Large Language Model) Agent code in the language of your choice. It provides a simple and intuitive syntax that allows developers to easily create prompts for AI agents without getting in the way of the creative process.
Key features of AgentFlow:
- Supports code generation for Python, TypeScript, and JavaScript (with more languages planned)
- Uses a .af file extension
- Employs an ultra-simple syntax focused on efficient prompting
This VSCode extension currently provides:
- Syntax highlighting for AgentFlow (.af) files
Note: A Language Server Protocol (LSP) implementation is planned for future releases to provide more advanced features.
AgentFlow uses a minimalist syntax:
-
Variables:
<!variable_name>Defines a variable which becomes a string input in code generation.
-
Titles:
.title This Is The Title Of The SectionThe rest of the line after
.titlebecomes the title. Code generation automatically converts the title to camelCase or snake_case based on language conventions.
[Provide installation instructions once the extension is published]
- Install the AgentFlow VSCode extension
- Create a new file with the
.afextension in VSCode - Start writing your AgentFlow prompts using the syntax described above
- Enjoy syntax highlighting for your AgentFlow files
- Initial release of AgentFlow syntax highlighting
- Language Server Protocol (LSP) implementation for enhanced functionality
- Code generation integration within VSCode
- Support for additional programming languages
If you'd like to contribute to the development of this extension or the AgentFlow language, please [provide information on how to contribute].
[Include your chosen license information here]
For more information on AgentFlow and its capabilities, please visit [your website or repository link].
Enjoy using AgentFlow!