pmax-asset-automation
Skill72 starsChanged 4 months agoMITView on GitHub
Audit and fix Performance Max campaign asset automation settings. Opts out of Google's auto-generated headlines, descriptions, videos, and image enhancements that often produce off-brand or hallucinated output. Auto-invoke when user says "audit PMAX asset automation", "check auto-created assets", "fix PMAX automatic assets", or "are PMAX campaigns opted out".
In this repository41 other files it gives its agents.
- account-diagnosticSkill
- ad-copy-generation-frameworkSkill
- ad-copy-verification-standardSkill
- add-account-negative-keywordsSkill
- ads-checkerSkill
- budget-guardianSkill
- budget-recommendation-calculatorSkill
- change-history-checkerSkill
- client-communication-standardsSkill
- competitor-analysis-v2Skill
- conversion-tracking-healthSkill
- dgen-automation-disableSkill
- fair-housing-complianceSkill
- ga4-campaign-cross-referenceSkill
- ga4-cross-analysisSkill
- ga4-lead-quality-investigationSkill
- gaql-query-patternsSkill
- geo-conflict-analyzerSkill
- google-ads-api-setupSkill
- google-ads-querySkill
- google-ads-samplesSkill
- impression-share-diagnosticsSkill
- investigation-methodologySkill
- lead-quality-pattern-analysisSkill
- lead-quality-recommendation-prioritizationSkill
- markdown-to-sheets-presenterSkill
- mcc-hack-auditSkill
- mutation-safetySkill
- neg-conflict-finderSkill
- non-serving-keyword-scannerSkill
- pmax-builderSkill
- portfolio-health-prioritizationSkill
- portfolio-pacing-rulesSkill
- rsa-bulk-editSkill
- rsa-refreshSkill
- rsa-single-accountSkill
- shared-budget-updaterSkill
- sqr-classifierSkill
- sqr-pipelineSkill
- underspending-investigationSkill
- youtube-placement-auditSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 92
Do not batch across accounts without a per-account dry run — PMax settings affect serving, and a bad portfolio-scale mutation is painful to …
- Distinctive: line 94
Before fixing anything unexplained, run rules.md's escalation ladder (attribute via change history first).
≈ “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.
Files
PMax Asset Automation
Audit Performance Max campaigns for Google's five asset-automation settings
and drive them to the house standard: all five OPTED_OUT. The audit is
shipped code (scripts/audit_pmax_asset_automation.py); the fix is a
documented mutation pattern run under
mutation-safety. Judgment calls (which flags matter,
when ON is legitimate, triage order) live in rules.md; exact
script mechanics live in
references/audit-contract.md.
Why This Exists
PMax campaigns ship with five asset automations ON by default: Google generates new text assets, expands your final URLs to pages you didn't pick, trims/remixes/creates YouTube videos, auto-crops images, and extracts new images from your landing pages — all served without per-asset approval. The generated assets frequently drift off-brand or contradict verified business claims, and the UI offers no portfolio-wide view (one campaign at a time, buried in Settings → Asset automation). For most serious advertisers the right default is all five off and manual asset management; this skill codifies that standard and scales the check across accounts.
What this skill deliberately does NOT do
- Never modifies an account. The shipped script is read-only; there is no fix script in this folder. Fixing is a mutation you run through the documented pattern below, gated by mutation-safety.
- Doesn't decide whether OPTED_IN is wrong. Some ONs are deliberate (client strategy, running experiments, evidence-backed exceptions) — rules.md owns that call and the coordinate-before-flipping cases.
- Doesn't see beyond the five tracked types. REMOVED campaigns are out of scope, and any automation type Google ships in the future is invisible until the script's list is extended (contract § fixed scope).
- Doesn't resolve account names. Except under
--all(which walks the MCC), accounts display asCID <n>— bring your own name lookup. - Writes nothing to disk. Console-only; redirect stdout when you need a record.
How to Check in the Google Ads UI (single campaign)
Campaign → Settings → expand Asset automation → all five toggles OFF. Fine for one campaign; useless at portfolio scale — that's what the script is for.
How to Audit via the API
Requires working Google Ads API access — google-ads-api-setup if you don't have it.
# Single account
python scripts/audit_pmax_asset_automation.py --cid 1234567890
# Multiple accounts (comma-separated)
python scripts/audit_pmax_asset_automation.py --cids "1234567890,2345678901"
# All accounts under the MCC (uses login_customer_id from your yaml)
python scripts/audit_pmax_asset_automation.py --all
# Credentials somewhere other than ./google-ads.yaml
python scripts/audit_pmax_asset_automation.py --all --config path/to/google-ads.yamlOutput: a per-account breakdown — every ENABLED or PAUSED PMax campaign,
its five settings with [OK]/[!] markers, per-campaign compliance
verdicts, and a portfolio summary. Read it in rules.md's triage order
(error lines first, count sanity second, findings third) — the summary
line All accounts fully compliant! has two known failure shapes
(examples.md 2 and 3). Full selection semantics, the
seeded-default mechanic, and every console shape:
references/audit-contract.md.
How to Fix via the API (Mutation)
Fixing is a mutation — it goes through mutation-safety. No exceptions.
- Dry run — show which campaigns will change and which settings go to
OPTED_OUT - User approves — explicit confirmation required
- Execute —
CampaignService.mutate_campaignswith an update onasset_automation_settings - Verify — re-run the audit ~24h later; settings have been observed reverting after campaign edits (rules.md false-alarm table)
Do not batch across accounts without a per-account dry run — PMax settings affect serving, and a bad portfolio-scale mutation is painful to unwind. Before fixing anything unexplained, run rules.md's escalation ladder (attribute via change history first).
Files
| File | Purpose |
SKILL.md | This file — workflow + routing |
rules.md | Judgment layer: risk ranking of the five settings, when-ON-is-legitimate, triage order, false alarms, escalation ladder, cadence |
examples.md | Three worked reads: the paused-campaign video call, the false "fully compliant", the all-five-ON baseline |
references/audit-contract.md | Line-derived script mechanics: selection, seeded defaults, error isolation, console shapes |
scripts/audit_pmax_asset_automation.py | The audit (read-only, console-only) |
After a Run
Nothing lands on disk — the report exists only in the console. A cold
session cannot tell an audit ran; redirect stdout
(> pmax-audit-$(date +%Y%m%d).txt) to keep a durable record, and always
capture the before/after pair around a fix. Account state never changes
from running the audit.
When to Load What
| Moment | Load |
| Reading any audit output — triage, risk ranking, is-this-ON-legitimate | rules.md |
| Exact selection/error/console-shape questions | references/audit-contract.md |
| First read, or the summary looks too clean | examples.md |
| Any fix — dry run, approval, execute | mutation-safety |
| An OPTED_IN nobody can explain — who flipped it, when | change-history-checker |
| Same problem, Demand Gen campaigns (ad-level automation) | dgen-automation-disable — the ad-level twin |
| Building a new PMax with the opt-outs pre-baked | pmax-builder — then audit after import |
| This check as one slice of a full account sweep | account-diagnostic — its automation checks route here to fix |
| The same flags surfaced by a creative-compliance audit | ads-checker — its automation checks (2, 6, 7) route here as the campaign-level fixer |
| No API credentials yet | google-ads-api-setup |
| The philosophy behind the standard (Empty > Inaccurate) | ad-copy-verification-standard |
Installation
mkdir -p .claude/skills/pmax-asset-automation/scripts .claude/skills/pmax-asset-automation/references
for f in SKILL.md README.md rules.md examples.md; do
curl -o .claude/skills/pmax-asset-automation/$f \
https://raw.githubusercontent.com/fourteenwm/ppc-ai-skills/main/pmax-asset-automation/$f
done
curl -o .claude/skills/pmax-asset-automation/references/audit-contract.md \
https://raw.githubusercontent.com/fourteenwm/ppc-ai-skills/main/pmax-asset-automation/references/audit-contract.md
curl -o .claude/skills/pmax-asset-automation/scripts/audit_pmax_asset_automation.py \
https://raw.githubusercontent.com/fourteenwm/ppc-ai-skills/main/pmax-asset-automation/scripts/audit_pmax_asset_automation.pyDoes 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
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.

