Headless DevOps Gives AI Agents Access to Delivery Workflows
DevOps.com, Monday, September 28th, 2026
Copado's Federico Larsen explains exposing DevOps platforms to AI agents via APIs, CLIs and MCP without losing controls.
Copado co-founder and CTO Federico Larsen tells DevOps.com's Alan Shimel that delivery platforms built around screens assume a human operator, which breaks down as developers delegate to coding agents in tools like Cursor and Claude Code, so Copado exposes capabilities through three layers: APIs, CLI commands, and packaged agent skills or MCP servers.
Broader agent access expands what teams must validate, including connected apps, IP ranges and agent behavior, with quality gates earlier in delivery.
Because agent output varies run to run, testing must check whether agents stay on task and follow requirements.