GitHub: AI Agents Can Now Create Fully Triaged Issues Without Human Intervention

GitHub Introduces Real-Time Duplicate Detection

Every active project on GitHub knows the problem: the same bug gets reported by five users in five different issues. GitHub now addresses this directly in the creation interface — while the user types a title and description, the system analyzes existing issues in the background and displays up to 3 suggestions for potential duplicates. This is the difference between manual searching after creation and automatic warning before submission.

What MCP Support for Issue Fields Brings

MCP (Model Context Protocol) — the standard enabling AI agents to access external tools and data — now supports reading and writing issue fields on the GitHub MCP server. This means AI agents can create issues and immediately set all metadata: labels, assignees, milestones, priorities, and custom fields. Triaging that an engineer would previously do manually, the agent handles automatically at the moment of issue creation.

Agentic Triaging as a New Standard of Work

The combination of these two features changes the workflow for development teams. Instead of issues accumulating untriaged, an agent can monitor inbound requests, detect duplicates, fill in all fields, and assign responsible parties — without any human intervention. GitHub thereby continues building infrastructure for agentic software development within its ecosystem.