ads-audit

Name: ads-audit
Author: AgriciDaniel

2.3kinstalls1.1kforks

How to install ads-audit?

agricidaniel/claude-ads/ads-audit

$npx -y skills add AgriciDaniel/claude-ads --skill ads-audit

Installs into the current project.

Summary

Run a source-grounded paid-advertising audit for one or more of Google, Meta, YouTube, LinkedIn, TikTok, Microsoft, Apple, Amazon, Reddit, Pinterest, Snapchat, and X. Use for full ad checks, account health reviews, paid-media diagnostics, partial audits after authentication or wo

Files · 1

View on GitHub

SKILL.md

1	# Paid Advertising Audit
2
3	Produce a versioned JSON audit bundle first, then render human deliverables from
4	that bundle. Never aggregate prose-only worker reports or claim coverage for a
5	platform whose required worker, sources, inputs, or controls are missing.
6
7	## Procedure
8
9	1. Read the main `ads` operating contract and thinking framework.
10	2. Create a run manifest with business context, date window, currency, timezone,
11	requested platforms, scopes, available data, and privacy classification.
12	3. Normalize exports, screenshots, manual metrics, or authenticated reads into an
13	account snapshot. Preserve source lineage and mark missing fields.
14	4. Discover active platforms. Confirm requested inactive or data-less platforms
15	rather than silently skipping them.
16	5. Load each selected platform capability manifest, control registry, dated source
17	entries, benchmarks, and applicable policy material.
18	6. Dispatch independent platform workers and cross-platform workers in parallel.
19	7. Validate every result against the common finding schema. Retry one transient
20	failure; record all other failures and recovery hints.
21	8. Run deterministic scoring. Do not calculate or repair scores in the prompt.
22	9. Synthesize systemic findings across measurement, budget, creative, landing
23	pages, experimentation, policy, and regulatory exposure.
24	10. Write one atomic run bundle and render the requested reports.
25	11. Verify bundle completeness, citations, privacy, and render integrity.
26
27	## Platform workers
28
29	Use a dedicated worker for every selected platform:
30
31	- `audit-google`
32	- `audit-meta`
33	- `audit-youtube`
34	- `audit-linkedin`
35	- `audit-tiktok`
36	- `audit-microsoft`
37	- `audit-apple`
38	- `audit-amazon`
39	- `audit-reddit`
40	- `audit-pinterest`
41	- `audit-snapchat`
42	- `audit-x`
43
44	Add cross-platform workers only when their inputs exist:
45
46	- Tracking and attribution.
47	- Creative and landing-page quality.
48	- Budget, pacing, and financial viability.
49	- Platform policy, privacy, and regulation.
50
51	## Required finding fields
52
53	Each worker returns conclusions, not files:
54
55	```json
56	{
57	"status": "ok",
58	"platform": "google",
59	"findings": [
60	{
61	"control_id": "G-EXAMPLE",
62	"result": "pass\|fail\|unknown\|not_applicable",
63	"severity": "critical\|high\|medium\|info",
64	"confidence": "high\|medium\|low\|none",
65	"source_classification": "evidence_based\|practitioner\|contested\|folklore",
66	"observation": "What the supplied data demonstrates",
67	"evidence_refs": ["input:...", "source:..."],
68	"recommendation": "Decision-complete next action or null"
69	}
70	],
71	"contradictions": [],
72	"missing_inputs": [],
73	"recovery_hints": []
74	}
75	```
76
77	Validate against the repository schema rather than relying on this illustrative
78	fragment when the installed schema is available.
79
80	## Completeness rules
81
82	- `complete`: every requested required worker returned valid results and every
83	scored platform meets normal evidence coverage.
84	- `provisional`: all required workers returned, but one or more platforms have
85	60-79% evidence coverage or stale non-critical evidence.
86	- `partial`: a required platform or cross-platform worker failed or was omitted.
87	- `insufficient_evidence`: a requested platform has less than 60% coverage.
88
89	Never substitute feature awareness for account health. Optional, beta, premium,
90	ineligible, or unavailable features belong in an opportunity list and are unscored.
91
92	For each optional or gated feature, check account, market, objective, and access
93	eligibility first. If unavailable or ineligible, record an `unscored_opportunity`
94	with the eligibility result and no health-score effect. Reject any request to
95	penalize health merely because a beta is unavailable.
96
97	## Required-worker failure and weighting
98
99	A failed authentication or worker does not stop analysis of independent successful
100	platforms, but it changes the whole bundle to `partial`. Record the failed platform,
101	missing evidence, recovery hint, and no platform health score. Exclude its weight
102	from portfolio health; never assign zero, preserve a stale historical weight, or
103	include it in the denominator. Renormalize weights only among successfully scored
104	comparable platforms. If defensible remaining weights are unavailable, withhold
105	portfolio health rather than inventing weights.
106
107	Example: when an all-platform audit succeeds except for Amazon authentication,
108	continue with the other platforms, mark Amazon failed/missing, exclude Amazon's
109	weight, label the bundle `partial`, and never call it complete.
110
111	## Synthesis boundaries
112
113	Separate t

People who liked this also viewed

6 picks

Type

ads-audit

2.3kinstalls1.1kforks

View on GitHub

How to install ads-audit?

agricidaniel/claude-ads/ads-audit

$npx -y skills add AgriciDaniel/claude-ads --skill ads-audit

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Paid Advertising Audit
2
3	Produce a versioned JSON audit bundle first, then render human deliverables from
4	that bundle. Never aggregate prose-only worker reports or claim coverage for a
5	platform whose required worker, sources, inputs, or controls are missing.
6
7	## Procedure
8
9	1. Read the main `ads` operating contract and thinking framework.
10	2. Create a run manifest with business context, date window, currency, timezone,
11	requested platforms, scopes, available data, and privacy classification.
12	3. Normalize exports, screenshots, manual metrics, or authenticated reads into an
13	account snapshot. Preserve source lineage and mark missing fields.
14	4. Discover active platforms. Confirm requested inactive or data-less platforms
15	rather than silently skipping them.
16	5. Load each selected platform capability manifest, control registry, dated source
17	entries, benchmarks, and applicable policy material.
18	6. Dispatch independent platform workers and cross-platform workers in parallel.
19	7. Validate every result against the common finding schema. Retry one transient
20	failure; record all other failures and recovery hints.
21	8. Run deterministic scoring. Do not calculate or repair scores in the prompt.
22	9. Synthesize systemic findings across measurement, budget, creative, landing
23	pages, experimentation, policy, and regulatory exposure.
24	10. Write one atomic run bundle and render the requested reports.
25	11. Verify bundle completeness, citations, privacy, and render integrity.
26
27	## Platform workers
28
29	Use a dedicated worker for every selected platform:
30
31	- `audit-google`
32	- `audit-meta`
33	- `audit-youtube`
34	- `audit-linkedin`
35	- `audit-tiktok`
36	- `audit-microsoft`
37	- `audit-apple`
38	- `audit-amazon`
39	- `audit-reddit`
40	- `audit-pinterest`
41	- `audit-snapchat`
42	- `audit-x`
43
44	Add cross-platform workers only when their inputs exist:
45
46	- Tracking and attribution.
47	- Creative and landing-page quality.
48	- Budget, pacing, and financial viability.
49	- Platform policy, privacy, and regulation.
50
51	## Required finding fields
52
53	Each worker returns conclusions, not files:
54
55	```json
56	{
57	"status": "ok",
58	"platform": "google",
59	"findings": [
60	{
61	"control_id": "G-EXAMPLE",
62	"result": "pass\|fail\|unknown\|not_applicable",
63	"severity": "critical\|high\|medium\|info",
64	"confidence": "high\|medium\|low\|none",
65	"source_classification": "evidence_based\|practitioner\|contested\|folklore",
66	"observation": "What the supplied data demonstrates",
67	"evidence_refs": ["input:...", "source:..."],
68	"recommendation": "Decision-complete next action or null"
69	}
70	],
71	"contradictions": [],
72	"missing_inputs": [],
73	"recovery_hints": []
74	}
75	```
76
77	Validate against the repository schema rather than relying on this illustrative
78	fragment when the installed schema is available.
79
80	## Completeness rules
81
82	- `complete`: every requested required worker returned valid results and every
83	scored platform meets normal evidence coverage.
84	- `provisional`: all required workers returned, but one or more platforms have
85	60-79% evidence coverage or stale non-critical evidence.
86	- `partial`: a required platform or cross-platform worker failed or was omitted.
87	- `insufficient_evidence`: a requested platform has less than 60% coverage.
88
89	Never substitute feature awareness for account health. Optional, beta, premium,
90	ineligible, or unavailable features belong in an opportunity list and are unscored.
91
92	For each optional or gated feature, check account, market, objective, and access
93	eligibility first. If unavailable or ineligible, record an `unscored_opportunity`
94	with the eligibility result and no health-score effect. Reject any request to
95	penalize health merely because a beta is unavailable.
96
97	## Required-worker failure and weighting
98
99	A failed authentication or worker does not stop analysis of independent successful
100	platforms, but it changes the whole bundle to `partial`. Record the failed platform,
101	missing evidence, recovery hint, and no platform health score. Exclude its weight
102	from portfolio health; never assign zero, preserve a stale historical weight, or
103	include it in the denominator. Renormalize weights only among successfully scored
104	comparable platforms. If defensible remaining weights are unavailable, withhold
105	portfolio health rather than inventing weights.
106
107	Example: when an all-platform audit succeeds except for Amazon authentication,
108	continue with the other platforms, mark Amazon failed/missing, exclude Amazon's
109	weight, label the bundle `partial`, and never call it complete.
110
111	## Synthesis boundaries
112
113	Separate t

People who liked this also viewed

6 picks

Type