Agentic DevOps Readiness

Summary of the resource kit shared by Mumshad Mannambeth regarding agentic AI in DevOps contexts.

Key Resources

90-Day Playbook: Governance-first path to running the first agentic AI workflow.
Agent Boundary Design: IAM templates for defining agent boundaries.
Eval Framework: 10 real CI failure scenarios (dependency conflicts, OOM kills, flaky tests, etc.) with pass/fail criteria and “what the agent should NOT do” rules.
Readiness Assessment: 4-layer stack assessment (Foundation, Platform, Agent Infrastructure, Agent Intelligence).
Tool Landscape: Reference for Claude Code, GitHub Copilot Agent, LangChain, Argo Workflows, kagent.

Vercel Breach Warning: Mention of the Vercel breach caused by an employee connecting a third-party AI tool (“Context”) to a corporate Google account with full read access.
Permission Matrix: Criticality of asking “what can this agent READ, WRITE, and EXECUTE?” before connection.