This page is a stub. I want an agentic tool use strategy that sits next to reliability and change management, built the same way: on models and values, then takeaways, then what you or your agent can produce.
Until it is written, start with AGENTS.md, Working with Claude, and the automation blueprint. The guardrails I already use: an agent that cannot reverse a change in 5 minutes needs a human, and agent output is a draft.