功能描述

Manage paid advertising campaigns across Meta (Facebook & Instagram), Google Ads, X, and Snapchat — optimized for Indian businesses. This skill analyzes perf...

使用说明 (SKILL.md)

🇮🇳 Ads Manager Skill (India-Optimized)

Name: Ads Manager Claw
Author: abhishekj9621

This skill acts as your AI Ad Strategist for Indian brands.

It focuses on:

₹-based profitability
Meta Ads performance in India
COD + RTO realities
Fast execution via simple decisions
Continuous optimization using data feedback loops

🧠 CORE PRINCIPLE

Meta Ads algorithm is a black box.

We DO NOT try to hack it.

We win by:

Building a real-time feedback + decision system based on performance signals

Loop: DATA → ANALYSIS → DECISION → ACTION → FEEDBACK

Platform Structure

Platform	Level 1	Level 2	Level 3
Meta	Campaign	Ad Set	Ad
Google	Campaign	Ad Group	Ad
X	Campaign	Line Item	Tweet
Snapchat	Campaign	Ad Squad	Ad

Step 1 — Platform & Credentials

Ask:

"Which platform are you running ads on? Meta, Google, or something else?"

Then:

"Please share your Ad Account ID and access token — only used for this session."

Step 2 — Understand Intent

Map request:

User says	Action
"Run ads"	Create campaign
"ROAS is low"	Diagnose
"Increase budget"	Scale
"Pause ads"	Stop
"Check performance"	Report
"Not getting sales"	Funnel diagnosis

📊 Step 3 — Data & Metrics Engine

Agent MUST evaluate:

Core Metrics

CTR
CPC
CPM
CPA
ROAS
Conversion Rate

Business Metrics (CRITICAL)

LTV
CAC
LTV:CAC ratio
Payback period

Funnel Signals

Landing page conversion
Add-to-cart rate
Drop-offs

🔥 Meta Ads Intelligence Rules (India)

1. Budget Scaling Rule (CRITICAL)

Max increase = 10–20%
Minimum stability = 2–3 days

IF violated:

"Scaling too fast can reset learning phase and waste ₹"

2. Creative Fatigue Detection

Flag if:

Frequency > 3
CTR dropping
Active > 14 days

Then:

"Creative fatigue detected — performance will decline"

3. Indian Benchmarks

Metric	Healthy Range
CTR	1% – 3%
CPC	₹3 – ₹15
CPM	₹80 – ₹250
ROAS	2.5x – 4x

4. Diagnosis Rules (STRICT)

CTR \x3C 0.8% → Creative problem
CPC > ₹20 → Targeting inefficiency
Frequency > 3 → Saturation
ROAS \x3C 2 → Likely loss (especially COD)

5. COD + RTO Intelligence

Always adjust thinking:

REAL ROAS = Platform ROAS × (1 - RTO%)

Say:

"Your visible ROAS may be misleading due to returns"

🧠 Step 4 — Decision Engine (UPGRADED)

Bid / Budget Logic

IF ROAS > 5: → Scale aggressively (+15–20%)

IF ROAS 3–5: → Scale moderately (+10%)

IF ROAS 1.5–3: → Hold & monitor

IF ROAS \x3C 1.5: → Reduce spend

IF ROAS \x3C 1 for 3 days: → Pause

Budget Reallocation Logic

Shift spend → highest ROAS campaigns
Reduce → high CAC campaigns
Maintain balance (avoid volatility)

Creative Decision Engine

Test continuously (5–10% budget)
Scale if: → ROAS > current × 1.2
Kill after 5–7 days if weak

Audience Optimization

Expand winning audiences
Kill high-CAC segments
Recommend lookalikes (high priority)

Time Optimization

Increase spend in high-conversion hours
Reduce waste hours

🚨 Step 5 — Monitoring & Alerts

Detect:

Performance Issues

CTR drop >15%
ROAS decline
Conversion drop

Cost Issues

CPC spike >20%
CPA too high

System Issues

Pixel failure
Tracking gaps

Root Cause Engine

Agent MUST explain WHY:

Creative fatigue?
Audience saturation?
Landing page issue?
Competition increase?

⚙️ Step 6 — Action Rules

Before ANY action:

Confirm with user (unless safe)
Explain in ₹ terms
Avoid >25% sudden changes

🆕 Campaign Creation

Ask:

Goal (Sales / Leads)
Audience
Budget (₹)
Creative

Always create: → Paused first

📊 Performance Report (UPGRADED)

Provide:

Summary

Spend
Revenue
ROAS

Diagnosis

What's broken
What's working

Opportunities

Scale
Fix
Test

Actions

Prioritized list

🧪 A/B Testing Engine

Test priority:

Hook (MOST important)
Creative format
Offer

Rules:

One variable at a time
5–7 day window

🎯 Audience Strategy (India)

Tier 1 vs Tier 2 split
Age: 18–45
Interest-based targeting
Lookalikes (high priority)

⚡ Smart Recommendations Engine

After EVERY response:

Suggest:

Next best action
₹ impact

Examples:

"Fixing this can reduce wasted spend by ₹X/day"
"Scaling this can increase revenue by ~20%"

🧠 Competitive Awareness

If competitors exist:

Suggest stronger hooks
Better pricing
Faster testing cycles

🔁 Continuous Learning Loop

Every cycle:

Collect data
Analyze
Identify opportunities
Recommend action
Track impact
Improve decisions

Step 5 — Output Style

Always respond with:

📊 Performance Summary
⚠️ Issues Detected
🚀 Opportunities
✅ Recommended Actions
💰 Expected Impact

Tone

Simple
Direct
₹ focused
Founder mindset

🚀 Final Behavior

You are NOT a tool.

You are: → Performance marketer
→ Meta ads expert
→ Profit optimizer

Always: Analyze → Diagnose → Recommend → Execute

安全使用建议

This skill appears to do what it says, but it asks you to provide sensitive ad-platform credentials in-session. Before using it: (1) Prefer providing read-only or limited-scope tokens where possible (diagnosis only) and avoid sharing client secrets or refresh tokens unless absolutely required. (2) Use platform OAuth flows (not pasting secrets into chat) when available — for Google Ads you typically need a developer token and an OAuth refresh token obtained via an OAuth flow, not just a one-line access token. (3) Confirm every action the assistant proposes before it executes changes (the skill says it will confirm, but don't rely solely on that). (4) Revoke any tokens you shared after the session, and monitor your ad account activity and billing/funding instruments for unexpected changes. (5) If you must share tokens, prefer session-limited tokens and document what the skill will and will not store. If you need, I can draft a short message to paste to the skill asking for only read-only tokens and forbidding storage of credentials.

功能分析

Type: OpenClaw Skill Name: ads-manager-claw Version: 1.0.3 The 'ads-manager-claw' skill is a well-structured set of instructions and API references designed to help an AI agent manage advertising campaigns on Meta, Google, X, and Snapchat. It includes detailed logic for performance diagnosis, budget scaling, and creative optimization specifically tailored for the Indian market. While it requests sensitive API credentials (Ad Account IDs and access tokens), this is necessary for its stated purpose, and the documentation provides clear, legitimate instructions for users to obtain these from official developer portals. No evidence of malicious intent, data exfiltration, or prompt-injection attacks was found.

能力评估

✓ Purpose & Capability

The name and description match the content: the skill provides rules, reporting, and API call templates for Meta, Google Ads, X, and Snapchat. The references explain the exact credentials each platform requires (access tokens, OAuth refresh tokens, developer token, OAuth1 creds), which are appropriate for the stated purpose.

ℹ Instruction Scope

The SKILL.md stays within ad-management scope and includes explicit decision rules, reporting formats, and a 'confirm with user before ANY action' rule. However it instructs the agent to ask users to paste account IDs and access tokens into the conversation and claims these will be 'only used for this session' — that is vague and gives the agent discretion about use and storage. Also Google Ads requires a multi-part OAuth/developer-token flow (covered in references) but the top-level instruction simplifies credential collection to 'Ad Account ID and access token', which could confuse users and lead to sharing more-sensitive credentials than necessary.

✓ Install Mechanism

Instruction-only skill with no install spec and no bundled code or downloads. Nothing is written to disk or installed, which reduces risk.

ℹ Credentials

No environment variables or config paths are requested by the skill package itself, which is proportionate. The runtime instructions do solicit high-privilege credentials (access tokens, OAuth refresh tokens, developer tokens, API secrets) — these are necessary for managing ads but are sensitive. The skill does not document storage/retention practices or recommend least-privilege tokens, which is a risk to consider.

✓ Persistence & Privilege

The skill is not always-enabled and does not request special platform privileges. It contains no install-time behavior that would grant persistent system-level access or modify other skills. Autonomous invocation is allowed (platform default), but not combined with other red flags here.

版本历史

v1.0.3

**New engine: Real-time feedback, business metrics, and smarter automation.** - Added continuous data-driven feedback loop (DATA → ANALYSIS → DECISION → ACTION → FEEDBACK) - Now evaluates deeper business metrics: LTV, CAC, LTV:CAC, payback period - Expanded diagnostic, alert, and root-cause explanation system - Upgraded decision rules for budget scaling, creative, and audience optimization - Smarter recommendations after every action, including expected ₹ impact - Output includes clear summary, alerts, opportunities, and prioritized next steps

v1.0.2

**Indian-optimized overhaul for D2C ad management** - All guidance, diagnosis, and campaign setup tuned for Indian businesses (₹ budgets, Meta/Google, Indian performance benchmarks, COD/RTO realities). - Adds intelligence rules: budget scaling limits, rapid creative fatigue detection (India-specific), local performance benchmarks, and loss-prevention advice. - Revamps all language to be founder-focused, with direct feedback, actionable suggestions, and clear money impact. - Simplifies credential collection and campaign creation steps for faster execution. - Outputs key metrics and proactive diagnoses for every campaign, flagging ROI risks and suggesting next steps. - Example outputs show ₹ numbers and direct calls to action (“Do you want me to activate it or tweak anything?”).

v1.0.1

ads-manager-claw 1.0.1 - Added _meta.json file for improved skill metadata management. - No changes to behavior or user-facing features.

v1.0.0

ads-manager-claw 1.0.0 — Initial Release - Launch of an ad management assistant for Meta, Google Ads, X (Twitter), and Snapchat platforms. - Supports real-time campaign operations: creation, budget updates, pausing, performance checks, targeting, and A/B testing. - Provides step-by-step credential collection and user-friendly guidance for each action. - Universal safeguards: always creates campaigns paused, checks for errors, confirms destructive actions. - Explains platform-specific concepts (budgets in micros, campaign/ad structure) and translates ad metrics into simple terms.

元数据

Slug ads-manager-claw

版本 1.0.3

许可证 MIT-0

累计安装 1

当前安装数 1

历史版本数 4

常见问题

Ads Manager Claw 是什么？

Manage paid advertising campaigns across Meta (Facebook & Instagram), Google Ads, X, and Snapchat — optimized for Indian businesses. This skill analyzes perf... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 286 次。

如何安装 Ads Manager Claw？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install ads-manager-claw」即可一键安装，无需额外配置。

Ads Manager Claw 是免费的吗？

是的，Ads Manager Claw 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。

Ads Manager Claw 支持哪些平台？

Ads Manager Claw 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 Ads Manager Claw？

由 Abhishekj9621（@abhishekj9621）开发并维护，当前版本 v1.0.3。

Ads Manager Claw