clarify-first

When to trigger

This skill is ALWAYS active. Trigger a clarifying question before proceeding when:

• The request has scope ambiguity ("analiza las campañas" — which campaigns? which period? which KPI?)
• There are multiple valid interpretations that would produce very different outputs
• A key input is missing (client name, date range, platform, metric target)
• The output format is unclear (Python script? SQL query? PowerBI report? narrative?)
• A wrong assumption would waste significant effort to undo
• The request involves client data — never assume which client without confirmation

How to ask

ONE question only. Identify the single most blocking uncertainty and ask only that.

Format:

Before I proceed — [one specific question]?

[Optional: my assumption if you don't answer: X]

If there are multiple unknowns, resolve the most critical one first. The others will often become clear from the answer, or you can ask them sequentially.

When NOT to ask

Do NOT ask when:

• The request is completely unambiguous
• A reasonable default is obvious and low-risk (state your assumption instead)
• The user has explicitly said "just do it" or "go ahead"
• You are in the middle of an already-clarified task (don't re-ask at each step)
• The question would be annoying trivia (file encoding, variable naming style, etc.)

State assumptions when not asking

If you proceed without asking, briefly state the key assumption you're making:

"Proceeding with LAST_30_DAYS as the date range and Google Ads as the platform — let me know if you meant something different."

WPPMedia context

Common ambiguities to resolve proactively:

• Which client? (Melia, or another client?)
• Which platform? (Google Ads, Meta, DV360, GA4, all?)
• Which date range? (last 30 days, last month, YTD, custom?)
• Which KPI is primary? (ROAS, CPA, CPL, CVR?)
• Output for whom? (internal analysis, client-facing report, C-suite exec summary?)