An AI coding agent keeps generating provider arguments that were deprecated or removed in the provider version the team uses. The team wants the agent to keep working with the current provider release. What should they add to the agent's toolchain?
Select an answer to reveal the explanation.
Short Explanation
Your agent is studying from an old textbook. The Terraform MCP server hands it this year's edition, straight from the registry, so it stops writing arguments that no longer exist. The trap is trying to fix stale knowledge with formatting or by going back in time to an old provider.
Full Explanation
Language models are trained on a snapshot of public code, so their knowledge of fast-moving provider schemas goes stale. The Terraform MCP server implements the Model Context Protocol so that an agent can call tools that search provider documentation, inspect module details and discover policies from the Terraform Registry at the time of the request. With it connected, the agent can look up the current resource arguments for the provider version in use before writing code, which addresses the root cause. Raising temperature only increases randomness in generation; it adds no new information and makes invented arguments more likely, not less. terraform fmt normalizes formatting and some syntax style; it does not know provider schemas and never renames or replaces provider arguments. Pinning an older major version would match the agent's stale knowledge but violates the stated goal of staying on the current release and forgoes fixes and features. Exam caveat: grounding the agent improves its drafts, but its output still needs terraform validate, plan review and policy checks like any other change. Operational check: ask the agent to generate a resource, confirm in its tool-call log that it queried registry documentation, and run terraform validate on the result to confirm no deprecation warnings.