agentleFS
gomarble-aigomarble-ai/marketing-agent/skills/google-pmax-pulse/SKILL.md

google-pmax-pulse

Skill15 starsChanged 37 days agoMITView on GitHub

PMax morning anomaly check — compares last 3 days vs prior 3 days across spend, conversions, and CPA/ROAS for every Performance Max campaign. Flags CPA spikes (over 25%), conversion drops (over 25%), and spend spikes without conversions. Read-only — no changes applied. Disciplined: does NOT recommend major changes unless the issue is severe, since PMax is a black box and overcorrection is the #1 killer.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the user didn't name an account, list their connected accounts and ask which one to use.

    line 11
  2. Distinctive:

    Do NOT suggest scaling on a 3d positive blip

    line 48
  3. Distinctive:

    Do NOT suggest budget cuts on a single-trigger alert

    line 49
  4. Distinctive:

    Do NOT suggest asset edits — PMax automatically rebalances

    line 50

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

PMax Pulse — 3-Day Anomaly Check (Read-Only)

For the Google Ads account the user names, scan every active Performance Max campaign and detect anomalies between the last 3 days and the prior 3 days. Read-only — never modify PMax campaigns from a daily pulse, the black-box dynamics mean knee-jerk reactions cost more than they fix.

If the user didn't name an account, list their connected accounts and ask which one to use.

Inputs (per PMax campaign, both 3d windows)

  • Spend
  • Conversions
  • CPA / ROAS

Anomaly triggers

SignalTrigger
CPA spike> 25% increase vs prior 3d
Conversion drop> 25% decrease vs prior 3d
Spend without conversionsSpend 3d up > 20%, conversions flat or down

Severity tiers

  • 🚨 Critical — multiple triggers fire on the same campaign AND the campaign represents > 15% of account spend. ONLY here should the recommendation include action language.
  • ⚠️ Alert — single trigger fires. Surface and watch.
  • 👀 Monitor — soft signal, < trigger threshold, but trend worth tracking.

Output format

Three sections, each a vertical list:

🚨 Critical (recommend immediate review)

  • [Campaign] — CPA ↑42%, conversions ↓30% over 3d. Spend $X (NN% of account). Consider pausing or moving to manual asset review.

⚠️ Alerts (surface, no action recommended yet)

  • [Campaign] — CPA ↑28% over 3d. Single signal. Watch for 3 more days before acting.

👀 Monitor (soft signals only)

  • [Campaign] — Spend ↑15% with conversions flat. Trend, not anomaly.

Disciplined rules

  • Do NOT suggest scaling on a 3d positive blip
  • Do NOT suggest budget cuts on a single-trigger alert
  • Do NOT suggest asset edits — PMax automatically rebalances
  • For Critical: state "consider pausing OR moving to manual asset review" — those are the only two safe responses to a true PMax crisis

End with: _All findings — no changes applied. Investigate Critical items first._

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.