Build a universal utility that transforms natural language tasks into precise command-line arguments.
You'll build an interactive command-line companion that turns plain-language requests into the exact arguments a tool expects. Ask it to undo your last commit and it will surface git reset --soft HEAD^, ready to run.
It runs as a universal wrapper you can call alongside your usual commands (for example, git+ drop the last commit) and always shows a clear preview for confirmation, so actions stay safe and deliberate.
Build a CLI that leverages LLM prompts
Implement robust argument parsing and path handling
Integrate rig-core with the tokio runtime
Create robust error handling with anyhow
Add friendly confirmations using dialoguer
Execute external commands and handle exit codes
Basic Rust syntax and ownership
Basic command-line experience
Familiarity with async/await concepts
Rust and cargo installed
OpenAI API key
Parsing inputs
Create a binary crate and configure anyhow for ergonomic errors.
Parse command-line args, extract the real command, and build natural_args.
Interacting with a model
Create a prompt template with role, task, placeholders, and constraints. Wire it to rig-core and tokio for async requests.
Running a command
Request user confirmation with dialoguer, then execute the composed command.
Run commands via tokio process, forward I/O, and handle exit codes.
Testing the agent
Build the release binary, install it to ~/.local/bin, and create an alias.
Set OPENAI_API_KEY, then run git+ to suggest precise git commands.
Hands-on guides for AI app development
