AI

AI CLI is a .NET-powered terminal assistant that exposes a set of AI-driven capabilities through a command-line interface. It can summarize text, generate release notes, create changelogs, produce code documentation, and perform actions that leverage MCP servers to automate tasks. Once installed, you run the ai command and supply tools and inputs as needed. For example, you can let AI summarize a long document, generate release notes for a project, or orchestrate actions across MCP-backed servers to automate workflows. The tool integrates OpenAI access via an API key, and can operate with extra options such as specifying tools, inputs, or providers to tailor the response to your environment.

Prerequisites:

• .NET SDK installed on your system (sufficient to support dotnet tool usage).
• An OpenAI API key (OPENAI_API_KEY) or other provider credentials as required by your setup.

Installation steps:

1. Install the AI CLI as a global .NET tool: dotnet tool install --global tryagi.ai --prerelease
2. Ensure the AI CLI is available in your PATH. On most systems, the tool will install as the ai command.
3. Set your API key and any needed environment variables, for example:
   • On Windows (PowerShell): $env:OPENAI_API_KEY = "<your_api_key>"
   • On macOS/Linux (bash/zsh): export OPENAI_API_KEY="<your_api_key>"
4. Verify installation: ai --version
5. Run with tools as needed, for example: ai --tools filesystem git ai --tools github[issues,labels] --input-file .prompt.md

Tips and common considerations:

• Ensure OPENAI_API_KEY (or equivalent) is set in your environment before invoking ai commands.
• When using MCP-backed actions, provide the appropriate inputs (directories, repository paths, tokens) and keep your credentials secure.
• If you encounter permission issues on macOS/Linux, ensure your shell profile exports PATH to include the dotnet tool binaries location.
• For GitHub-related tasks, you may need a GITHUB_TOKEN with the required scopes (read/write access to issues/labels) and to supply it via the environment.
• The AI CLI can be used to orchestrate MCP servers for automated workflows; check your MCP server definitions to confirm available commands and inputs.
• If you plan to run this in CI, ensure the runner has access to the necessary environment variables and network access to OpenAI endpoints.