account-audit
Skill2 starsChanged 7 months agoMITView on GitHub
SKILL __Account Audit__ — Pull all performance data from a Meta Ads account, structure it into navigable dashboards and queryable data files. Covers: account overview, campaign breakdown, audience analysis (age/gender/placement/time), creative performance (image vs video, copy angles, hook rate), and A+ Advantage+ settings audit. Use this skill whenever the user mentions: 'audit', 'account audit', 'analizza account', 'analisi account', 'account review', 'performance review', 'cosa non va nell account', 'diagnosi account', 'health check', 'audit Meta', 'audit Facebook Ads', 'analisi campagne', 'panoramica account', 'account overview', or any request to understand what's happening across an entire ad account.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- Murtadha-Najem/instagram-account-audittheir own version, first seen Oct 15, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- Murtadha-Najem/instagram-account-audit★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository3 other files it gives its agents.
- skills/single-product-campaign/CLAUDE.mdCLAUDE.md
- single-product-campaignSkill
- winner-scalerSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 11 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
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 162
Create a React dashboard (.jsx) with tabs:
- Distinctive: line 239
Ask user for the two periods.
- Distinctive: line 273
Confidence note: If total purchases < 50, add: "⚠️ Low volume — {N} total purchases.
- Distinctive: line 296
No "images work better" conclusion: If one format has 3× the spend of the other, the comparison is skewed — flag it.
- Distinctive: line 332
If ≥3 signals trend negative → "Multiple signals suggest potential saturation.
- Distinctive: line 334
If signals are mixed → "No clear saturation pattern.
- Distinctive: line 347
No opinion on whether it's "enough." Just the numbers with comparison across periods if available.
- Distinctive: line 364
Every interrogation template follows these rules:
- Distinctive: line 367
Spend imbalance: If comparing two groups and one has 3×+ the spend, flag: "⚠️ Uneven spend — {X}€ vs {Y}€.
- Distinctive: line 369
No superlatives without context: Never say "best" or "worst" without showing the volume.
- Distinctive: line 373
All data is pulled via Pipeboard Meta MCP tools, not Ads Manager exports
≈ “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
Account Audit
Pull all performance data from a Meta Ads account, organize it into dashboards for visual analysis and structured data files for follow-up interrogation with Claude.
What it does
Takes an ad account and a date range, pulls everything, and produces:
- Interactive dashboard — React with tabs for each analysis dimension
- Structured data pack — CSV files saved to disk, one per dimension, ready to be loaded back into context for Q&A
The skill does NOT produce a diagnosis. The data is the deliverable. Diagnosis happens in the conversation after, when the user asks questions and Claude has the structured data to answer from.
What I need from you
- Ad account ID — e.g.
act_1240474954407657 - Date range — Period to analyze (e.g. "last_90d", "2025-09-01 to 2026-03-31")
- Comparison period (optional) — For trend analysis (e.g. "compare Sep-Nov vs Dec-Feb")
- Purchase event name (optional) — Default: "purchase". Some accounts use custom events.
I'll automatically look up the account's page, Instagram, and pixel.
Workflow
Step 0 — Account setup
Tool: get_account_info
→ Extract: account name, currency, timezone
Tool: get_account_pages
→ Extract: page_id, page_name
Tool: get_instagram_accounts
→ Extract: instagram_account_id
Tool: get_pixels
→ Extract: pixel_idSave as account_meta.json.
Step 1 — Account overview
Tool: get_insights
Level: account
Fields: spend, impressions, reach, clicks, ctr, cpc, actions, cost_per_action_type, purchase_roas
Date preset: {date_range}
Breakdowns: none (aggregate)Also pull with time increment (daily or monthly depending on range length) for trend line.
Output: 01_account_overview.csv — columns: date, spend, impressions, reach, clicks, ctr, cpc, purchases, cpa, roas
Step 2 — Campaign breakdown
Tool: get_campaigns
→ All campaigns (ACTIVE + PAUSED)
Tool: get_insights
Level: campaign
Fields: campaign_id, campaign_name, objective, spend, impressions, clicks, ctr, actions, cost_per_action_type, purchase_roas
Date preset: {date_range}Output: 02_campaigns.csv — columns: campaign_id, campaign_name, objective, status, spend, impressions, clicks, ctr, purchases, cpa, roas
Step 3 — Ad set breakdown
Tool: get_insights
Level: adset
Fields: adset_id, adset_name, campaign_id, campaign_name, spend, impressions, clicks, actions, cost_per_action_type, daily_budget, optimization_goal
Date preset: {date_range}Output: 03_adsets.csv — columns: adset_id, adset_name, campaign_id, campaign_name, spend, purchases, cpa, daily_budget, optimization_goal
Step 4 — Audience breakdowns
Pull account-level insights with demographic breakdowns:
Tool: get_insights (×4 calls)
Level: account (or campaign if user wants per-campaign)
Date preset: {date_range}
Call A — Breakdown: age, gender
Call B — Breakdown: publisher_platform, platform_position
Call C — Breakdown: hourly_stats_aggregated_by_advertiser_time_zone
Call D — Breakdown: (monthly or weekly time increment for trend)Output files:
04a_age_gender.csv— columns: age, gender, spend, impressions, clicks, purchases, cpa, roas04b_placements.csv— columns: platform, position, spend, impressions, clicks, purchases, cpa, roas04c_hourly.csv— columns: hour, spend, impressions, clicks, purchases, cpa, roas04d_trend.csv— columns: period, spend, impressions, clicks, purchases, cpa, roas
Step 5 — Creative performance
Tool: get_ads
→ All ads in the account for the period
Tool: get_insights
Level: ad
Fields: ad_id, ad_name, adset_id, adset_name, campaign_id, campaign_name, spend, impressions, clicks, ctr, cpc, actions, cost_per_action_type, purchase_roas
Date preset: {date_range}
Tool: get_ad_creatives (for each ad, or bulk)
→ Extract: creative_id, image_hash/video_id, thumbnail_url, object_story_spec (title, body, description, call_to_action, link)For each creative, extract and flatten:
- Format: image or video (infer from whether video_id is present)
- Headline: from object_story_spec → link_data → title
- Primary text: from object_story_spec → link_data → message
- Description: from object_story_spec → link_data → description
- CTA type: from call_to_action → type
- Creative ID: to detect same creative used across multiple ads
- A+ settings: asset_feed_spec or degrees_of_freedom_spec → advantage_plus creative enhancements, product_extensions (Show Products) OPT_IN/OPT_OUT
Copy angle tagging: Group ads by primary text content. Assign a short label to each distinct copy angle (e.g. "outlet discount", "free shipping", "brand story"). Count how many unique angles exist per time period to detect copy diversity collapse.
Output files:
05a_ads_performance.csv— columns: ad_id, ad_name, adset_id, adset_name, campaign_id, campaign_name, creative_id, format, spend, impressions, clicks, ctr, cpc, purchases, cpa, roas05b_creatives_detail.csv— columns: creative_id, ad_id, format, headline, primary_text, description, cta_type, advantage_plus_settings, product_extensions05c_copy_angles.csv— columns: copy_angle_label, primary_text_sample, ad_count, total_spend, purchases, cpa
Step 6 — Save data pack
Save all CSV files to /home/claude/audit_{account_id}/:
audit_act_XXXXX/
├── account_meta.json
├── 01_account_overview.csv
├── 02_campaigns.csv
├── 03_adsets.csv
├── 04a_age_gender.csv
├── 04b_placements.csv
├── 04c_hourly.csv
├── 04d_trend.csv
├── 05a_ads_performance.csv
├── 05b_creatives_detail.csv
└── 05c_copy_angles.csvThese files stay on disk so Claude can reload them in follow-up questions.
Step 7 — Build dashboard
Create a React dashboard (.jsx) with tabs:
Tab 1 — Overview
- Total spend, purchases, CPA, ROAS as KPI cards
- Trend line (spend + CPA over time)
Tab 2 — Campaigns
- Table: all campaigns sorted by spend, with purchases, CPA, ROAS
- Highlight: campaigns with zero purchases, campaigns with CPA > 2× account average
Tab 3 — Audience
- Age × Gender heatmap (CPA as color)
- Placement breakdown bar chart
- Hourly performance chart (ROAS or CPA by hour)
Tab 4 — Creatives
- Table: all ads sorted by spend, with format, CPA, creative_id
- Flag: same creative_id appearing in multiple campaigns (with CPA comparison)
- Format split: image vs video aggregate performance
Tab 5 — Copy & A+
- Copy angles table: angle label, ad count, spend, CPA
- Copy diversity over time (number of distinct angles per month)
- A+ settings breakdown: OPT_IN vs OPT_OUT counts and avg CPA per setting
Save dashboard to output directory.
Step 8 — Interrogation menu
Present the dashboard, then show the user an actionable menu. Each option is a pre-built analysis with specific files to load and a delta-based presentation — no speculation, only measurable comparisons.
Dashboard ready. Data pack saved with {N} CSV files.
What do you want to dig into?
📊 PERFORMANCE
1. Where is my budget going? (spend distribution)
2. What changed between period A and period B?
3. Which campaigns have zero or near-zero conversions?
👥 AUDIENCE
4. Who is my best customer? (age/gender/placement)
5. When should I be spending? (hourly/daily patterns)
🎨 CREATIVE
6. Image vs video — which format converts better?
7. Am I reusing the same creatives too much?
8. How diverse is my copy? (angle analysis over time)
🔍 HEALTH CHECK
9. Saturation check
10. Testing velocity check
11. A+ settings impact check
Just give me the number.Interrogation templates
Each menu option has a defined template: which files to load, what to compute, how to present. Claude follows these templates exactly — no conclusions, only deltas and comparisons.
1. Where is my budget going?
Files: 02_campaigns.csv
Present:
- Campaigns ranked by spend (descending)
- % of total spend per campaign
- For each campaign: spend, purchases, CPA
- Highlight: campaigns with >10% of total spend and zero purchases
2. What changed between period A and period B?
Files: 04d_trend.csv, 05a_ads_performance.csv, 05c_copy_angles.csv, 05b_creatives_detail.csv
Ask user for the two periods. Then compute and show side by side:
| Metric | Period A | Period B | Delta |
| Daily spend | |||
| Number of active ads | |||
| Number of distinct creatives | |||
| Number of copy angles | |||
| Frequency (if available) | |||
| % budget on ads older than 14 days | |||
| A+ OPT_IN vs OPT_OUT split | |||
| CPA | |||
| Purchases |
No "because" statements. Present the table and let the user see what moved together.
3. Which campaigns have zero or near-zero conversions?
Files: 02_campaigns.csv
Present:
- Filter: campaigns with spend > 0 and purchases ≤ 1
- Table: campaign name, spend, impressions, clicks, purchases
- Total spend wasted on zero-conversion campaigns
4. Who is my best customer?
Files: 04a_age_gender.csv, 04b_placements.csv
Present:
- Age × gender table sorted by CPA (lowest first)
- Include: spend, purchases, CPA, % of total purchases per segment
- Minimum threshold: only show segments with ≥ 3 purchases (below that, not statistically meaningful — flag it)
- Same for placements: sorted by CPA, with volume threshold
Confidence note: If total purchases < 50, add: "⚠️ Low volume — {N} total purchases. These patterns may not hold with more data."
5. When should I be spending?
Files: 04c_hourly.csv
Present:
- Hourly chart: spend and CPA per hour
- Highlight: top 3 hours by CPA (lowest) and top 3 by spend
- Gap analysis: hours where CPA is low but spend is also low (potential opportunity)
- Hours where CPA is high and spend is high (potential waste)
Confidence note: Same volume threshold. If purchases per hour < 3, that hour's CPA is unreliable.
6. Image vs video performance
Files: 05a_ads_performance.csv
Present:
- Aggregate: image total spend/purchases/CPA vs video total spend/purchases/CPA
- Top 5 ads by CPA for each format
- Volume context: "Images: {N} ads, {X} total spend. Video: {N} ads, {X} total spend."
No "images work better" conclusion. If one format has 3× the spend of the other, the comparison is skewed — flag it.
7. Am I reusing creatives too much?
Files: 05a_ads_performance.csv, 05b_creatives_detail.csv
Present:
- Find creative_ids that appear in multiple ads/campaigns
- For each duplicate: show CPA per ad/campaign where it appears
- Table: creative_id, number of ads using it, campaigns, CPA range (min-max)
- Age of each creative (first seen vs now, if time data available)
8. How diverse is my copy?
Files: 05c_copy_angles.csv, 04d_trend.csv
Present:
- Number of distinct copy angles per month/period
- Trend: is it increasing, stable, or decreasing?
- Table: angle label, ad count, spend, CPA
- % of total spend concentrated on the top 1 and top 3 angles
9. Saturation check
Files: 04d_trend.csv, 05a_ads_performance.csv
Present a panel of signals, no conclusion:
| Signal | Value | Trend |
| CPA (weekly) | current vs 4 weeks ago | ↑ ↓ → |
| Frequency (if available) | current vs 4 weeks ago | ↑ ↓ → |
| % spend on ads > 14 days old | ||
| Number of new creatives last 2 weeks | ||
| Same creative_id best CPA: first week vs now |
If ≥3 signals trend negative → "Multiple signals suggest potential saturation. Consider testing new creatives and audiences."
If signals are mixed → "No clear saturation pattern. Individual signals below for your assessment."
This is the ONE place where Claude can suggest, but only with ≥3 converging signals, and framed as "consider" not "you have."
10. Testing velocity check
Files: 05a_ads_performance.csv, 04d_trend.csv
Present:
- Number of new ads launched per week/month (based on first date with spend)
- % of total budget on ads with < 7 days of data
- Number of ads that received < spend_threshold (from winner-scaler) and were paused — "killed before tested"
No opinion on whether it's "enough." Just the numbers with comparison across periods if available.
11. A+ settings impact check
Files: 05b_creatives_detail.csv, 05a_ads_performance.csv
Present:
- Split ads by A+ setting (OPT_IN vs OPT_OUT) for each feature (enhancements, product_extensions)
- For each split: count of ads, total spend, avg CPA
- Table format, no recommendation
Caveat always included: "This is a correlation, not an A/B test. Ads with different A+ settings may also differ in creative, audience, and budget."
Confidence framework
Every interrogation template follows these rules:
- Volume thresholds: Any segment with < 3 conversions gets flagged as "⚠️ Low volume — treat with caution"
- Spend imbalance: If comparing two groups and one has 3×+ the spend, flag: "⚠️ Uneven spend — {X}€ vs {Y}€. Comparison may be skewed."
- Correlation ≠ causation: Any time two metrics move together, never say one caused the other. Say "these moved together" and list both.
- No superlatives without context: Never say "best" or "worst" without showing the volume. "Lowest CPA" always comes with "on {N} purchases and {X}€ spend."
Important notes
- All data is pulled via Pipeboard Meta MCP tools, not Ads Manager exports
- Currency and timezone come from the account settings, never assume EUR
- The dashboard hardcodes data from the CSVs — if the user wants to refresh, re-run the skill
- For accounts with 100+ ads, the creative detail step may require multiple API calls — batch where possible using
bulk_get_ad_creativesandbulk_get_insights - The skill is read-only except for pulling data — it never creates, modifies, or pauses anything
Example
Account: act_1240474954407657
Date range: September 1, 2025 to March 31, 2026
Compare: Sep-Nov vs Dec-Feb vs MarOutput: 5-tab dashboard + 10 CSV files covering 6 months of performance data, ready for interrogation.
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
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.

