agentleFS
EricStrohmaierEricStrohmaier/adchat-mcp/skills/scale-winners/SKILL.md

scale-winners

Skill1 starsChanged 3 days agoMITView on GitHub

Grow spend on Meta or Google Ads campaigns, ad sets or ads that are clearly working, without breaking them. Use when the user asks to scale, increase budget, "put more money behind what works", find winners, or move budget from losers to winners in Meta (Facebook/Instagram) or Google Ads through AdChat.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    All of these, over the last 7 to 14 days:

    line 14
  2. Distinctive:

    No more than +20% per object per 3 days unless the user insists, and say the risk.

    line 61
  3. Very specific:

    Don't scale an ad set in Meta's learning phase or with delivery issues (meta_diagnose, kind: delivery_issues).

    line 65
  4. Very specific:

    Never raise budget to "fix" a drop.

    line 67

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Scale winners

Goal: put more money behind what is reliably beating the user's target, in steps small enough that performance holds. Scaling too fast is the most common way to turn a winner into a loser.

What counts as a winner

All of these, over the last 7 to 14 days:

  • Cost per result at or below target (or ROAS at or above target). Get the target from get_account_briefing; if there is none, ask.
  • Enough volume to trust: roughly 15+ conversions in the window.
  • Stable: not one lucky day. Check daily numbers.
  • Room to grow: on Meta, frequency under ~3 for prospecting; on Google, impression share lost to budget (not to rank).

Steps

  1. Context. get_account_briefing (targets, monthly budget limits) and the user's spending rules if they mention them.
  2. Find candidates.
    • Meta: meta_insights at level: adset (or campaign with campaign budgets), date_preset: last_14d, then time_increment: 1 on the top ones for stability.
    • Google: google_list with level: campaigns for the last 14 days. For budget-limited campaigns, google_run_gaql with metrics.search_budget_lost_impression_share (money in micros there).
  3. Check profit. get_true_profit if revenue is connected: scale on real profit, not platform-reported ROAS alone.
  4. Pick the move.
    • Raise budget 15–20% on the winner. Wait 3 to 4 days before the next step.
    • Move budget from a clear loser to the winner when the total must stay the same (pair this with cut-wasted-spend).
    • Duplicate a winning Meta ad set into a new audience with meta_duplicate (level: adset); the copy starts PAUSED.
  5. Propose. Show a table: object, current budget, new budget, why (numbers).
  6. Apply after Yes. Budget increases always need the user's Yes (except trusted mode). Meta: meta_update with level: adset or campaign and the new daily_budget in cents; several at once: meta_bulk_update with budget_change_percent and preview: true first. Google: google_update_campaign_budget with the new daily_budget in currency units.
  7. Follow up. Suggest checking again in 3 to 4 days before another step.

What to show the user

Scale proposal · Acme Shop · last 14 days · target CPA $25

Ad set               Spend   Purch.  CPA     Freq  Budget now → new
Lookalike 1% US      $812    52      $15.60  1.9   $60.00 → $72.00 (+20%)
Retargeting 30d      $301    21      $14.30  3.6   hold (frequency high)

Guardrails

  • No more than +20% per object per 3 days unless the user insists, and say the risk. The user's spending rules may cap it anyway; a refusal comes back with reasons.
  • Meta budgets in cents (7200 = $72.00). Google budgets in currency units (72.00).
  • A Google shared budget changes every campaign that uses it: say so.
  • Don't scale an ad set in Meta's learning phase or with delivery issues (meta_diagnose, kind: delivery_issues).
  • Never raise budget to "fix" a drop. Diagnose first (diagnose-performance).

Example prompt

"Which of my Facebook ad sets should get more budget? Go ahead with the best one."

Briefing → ad set insights 14 days → daily check on the top 3 → true profit → proposal table → "Raise Lookalike 1% US from $60 to $72 a day?" → Yes → meta_update daily_budget: 7200.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.