GitHub Next introduced Agentic Workflows, a research project that embeds autonomous AI agents into GitHub Actions. Unlike traditional deterministic workflows, these combine YAML configuration with natural language instructions, allowing AI to interpret context and adapt execution dynamically. The workflows support multiple AI engines (GitHub Copilot CLI, Claude, OpenAI), maintain security through minimal permissions and sandboxed execution, and extend existing Actions rather than replacing them. A practical example demonstrates how an agent automatically fixed a Terraform provider upgrade issue by analyzing CI logs, updating code syntax, and pushing changes—all triggered by a simple comment.

Table of contents
What are Github Agentic Workflows?Why I think this mattersPrerequisitesInstalling and Running Your First Agentic WorkflowReal-World Example: Terraform Provider Upgrade FixSecurity ConsiderationsKey TakeawaysAdditional readingSort: