A GitHub agentic workflow

GitHub agentic workflows combine standard GitHub Actions with AI agents to handle unstructured or semi-structured data. The author describes a real-world use case: automating the parsing of product release notes to generate upgrade analysis config files, a task previously done manually. Key steps covered include initializing workflows via CLI or GUI, writing workflows in Markdown and compiling them to YAML, and running them with a special Copilot token. Practical issues encountered include forgetting to compile Markdown to YAML before pushing, permission problems with auto-compilation workflows, Windows/Linux line-ending mismatches, and the inability to use GitHub Marketplace actions inside agentic workflows. The system prompt used for the agent is also shared, highlighting security constraints and prompt injection defenses.