GitHub Copilot's new feature could be a Devin killer (github.blog)

GitHub announced "Agent HQ," a platform-level integration that makes coding agents a native part of the GitHub workflow and the Copilot experience. Over the coming months, third‑party coding agents from Anthropic, OpenAI, Google, Cognition, xAI and others will appear inside GitHub under paid Copilot subscriptions; Copilot Pro+ users can already try OpenAI Codex in VS Code Insiders. Core features include a persistent "mission control" UI across GitHub, VS Code, mobile and the CLI to assign, steer and monitor parallel agents; Plan Mode to capture project context and generate step‑by‑step plans; AGENTS.md for source‑controlled agent rules and guardrails; and an MCP Registry in VS Code to discover and install specialist MCP servers. GitHub also adds branch controls, identity features for agent attribution, and a control plane for governance, audit logging and access policies. Technically and organizationally, Agent HQ aims to unify a fragmented agent ecosystem into a single orchestration layer that preserves Git primitives (branches, PRs, Actions) while adding enterprise controls. Public previews ship for GitHub Code Quality—extending Copilot’s checks to maintainability and reliability—and the Copilot metrics dashboard for org‑level usage visibility. For teams this promises easier parallel automation, consistent agent behavior, and stronger compliance and observability; for the broader AI tooling market, it could reduce friction for adopting multi‑vendor agents and displace standalone agent management solutions.