New resource: pipefy_ai_automation

[Danielmoraisg]

Summary

AI automations are automations whose action is generate_with_ai: a prompt plus the output fields it fills. Keeping prompts in git and reviewing prompt changes through terraform plan is a natural fit. The API covers the full lifecycle: createAutomation with the generate_with_ai action, updateAutomation, deleteAutomation, and validateAiAutomationPrompt as a preflight.

Proposed schema

• id (computed)
• pipe_id (required, RequiresReplace)
• name (required)
• prompt (required)
• field_ids (required, list: the output fields the AI fills)
• event_id (optional: trigger; defaults to the API default when unset)
• active (optional bool)

Notes

• pipefy_automation can already create these generically through action_params; the dedicated resource adds typed attributes and prompt validation.
• Call validateAiAutomationPrompt before create and update and surface its findings as diagnostics (it checks prompt field references, output field ids, event validity, AI enablement on the pipe, and credit budget).
• Read filters on action_id = generate_with_ai; an id pointing at a regular automation should be a clear error, not silent drift.