Examples
Yesware supports prospecting work where speed matters but an unreviewed contact, signal, or message can create more risk than value.
Operating contexts
Test a market thesis against recognizable companies without widening the ICP after seeing the output. Preserve target geography, stage proxies, exclusions, event dates, and the reason each account entered the list.
Review question: Would this remain a fit if the most attractive public signal disappeared?
Resolve company identity before changing routing fields. Track source, transformation, fallback order, and a visible unknown state when providers disagree or a domain changes hands.
Review question: Can an operator explain and reverse every CRM update?
Prepare brief, evidence-led account context for a human-reviewed draft. Keep role relevance, email verification, suppression, regional policy, and the decision to send in separate checks.
Review question: Can the rep identify fact, inference, and unsupported personalization?
GTM engineers can use the same record to inspect package permissions, rate limits, output schema, secret handling, and deletion paths. Their acceptance test is different: can the workflow fail safely without silently accepting a weak match or exposing a credential? Keeping that test adjacent to the business test makes implementation choices visible to the people accountable for their consequences.
These are controls rather than performance statistics. They offer a common language for reviewing waterfall enrichment, contact data, buying signals, message relevance, and email deliverability without promising coverage, reply rate, or inbox placement. A control can be challenged, documented, and repeated; an unsupported benchmark cannot.
Copy the command, run it in a controlled environment, and use the acceptance question for your role to judge the first output.
npx -y @okki-global/okki-go-taroball