Configure prompt templates and account scope

Starting conditions and inputs

Choose prompt or dashboard instructions and their target scope.

Values for this task

entry_url, template_id, kind, account_ids, instruction_text, prompt_body

Command sequence

open "$entry_url" && snapshot -i

click "$prompt_row" && snapshot -i

get text body

fill "$prompt_body_input" "$prompt_body" && press 'Tab' && snapshot -i

reload && snapshot -i

get value "$prompt_body_input"

Ref targets and expected results

#	Command / ref	Target and verification
1	`click` `$prompt_row`	Target prompt template record
2	`get text`	Read organization scope and intended use.
3	`fill` `$prompt_body_input`	Prompt template body input
4	`press`	Leave the body input.
5	`reload`	Reload the same prompt template.
6	`get value` `$prompt_body_input`	Compare the saved body with the expected value.

Check the result

Values persist and the change is ready for response review.

Resuming after interruption

Read the same IDs again. Preserve correct settings and update only differences.

Choose the prompt kind

Use prompt for customer-channel AI instructions and dashboard for operator AI instructions. Distinguish these from outbound message templates. Decide which AI behavior you intend to change.

Set organization and account scope

Check the template organization and account scope. Keep account-specific instructions out of shared scope. Test representative questions after changing the target to confirm which instructions apply.

Specify answer scope, unknown-information handling, staff follow-up, and ending conditions. Reuse reviewed Entity facts. For example, require an availability check instead of guessing reservation times.

Check tools and permissions

Check whether the requested behavior is supported by actual tools and permissions. A prompt cannot grant data access. Test that unauthorized requests are rejected or referred to the responsible staff member.