技能详情（站内镜像，无评论）
作者：danyangliu @danyangliu-sandwichlab
许可证：MIT-0
MIT-0 ·免费使用、修改和重新分发。无需归因。
版本：v1.0.0
统计：⭐ 0 · 210 · 0 current installs · 0 all-time installs
⭐ 0
安装量（当前） 0
🛡 VirusTotal ：良性 · OpenClaw ：良性
Package：danyangliu-sandwichlab/creative-ads-helper
安全扫描（ClawHub）
- VirusTotal ：良性
- OpenClaw ：良性
OpenClaw 评估
The skill is an instruction-only ad-creative assistant whose declared inputs and runtime guidance align with its stated purpose and it doesn't request extra credentials, installs, or system access.
目的
The name, description, and required/optional input contract (product_offer, target_audience, funnel_stage, existing_creative_context, optional performance_data, brand_guidelines, etc.) are consistent with generating and scoring ad creatives across platforms. There are no unexpected credential or cloud requirements for the described functionality.
说明范围
SKILL.md provides concrete workflows, decision rules, rubrics, and examples confined to creative generation, scoring, and fatigue detection. It does not instruct reading system files, environment variables, or contacting external endpoints. 'Route to Ads Compliance Review' implies human escalation rather than an automated external call.
安装机制
No install spec and no code files — this is instruction-only. Nothing is downloaded or written to disk by the skill itself, which minimizes supply-chain risk.
证书
The skill declares no required environment variables, credentials, or config paths. That is proportionate to its stated purpose; it only needs user-provided creative and performance data.
持久
The skill does not request always:true, does not modify other skills, and does not require persistent agent privileges. Autonomous invocation is allowed by platform default but the skill does not add extra persistence or broad privileges.
综合结论
This skill appears coherent and low-risk because it is instruction-only and requests no credentials or installs. Before using it, avoid pasting sensitive PII or account credentials into the input fields (performance_data should be de-identified if it contains user data). Review any ad copy the skill produces for policy compliance, legal claims, and brand alignment before publishing; automated scoring and heuristics are helpful but not a substi…
安装（复制给龙虾 AI）
将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。
请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「Creative Helper」。简介：Generate and evaluate ad creatives for Meta (Facebook/Instagram), TikTok Ads, Y…。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/danyangliu-sandwichlab/creative-ads-helper/SKILL.md
（来源：yingzhi8.cn 技能库）SKILL.md
---
name: creative-ads-helper
description: Generate and evaluate ad creatives for Meta (Facebook/Instagram), TikTok Ads, YouTube Ads, Google Ads, and Amazon Ads with script generation, performance scoring, fatigue timing, and inspiration references.
---
# Creative Helper
## Purpose
Core mission:
- Generate image and video ad script concepts.
- Analyze post-launch creative performance and score assets.
- Detect fatigue cycles and recommend replacement timing.
- Provide inspiration references and industry creative patterns.
## When To Trigger
Use this skill when the user asks for:
- new ad concept generation
- creative scoring and post-analysis
- fatigue diagnosis and replacement schedule
- vertical-specific inspiration examples
High-signal keywords:
- creative, ads, script, video, image
- performance, score, fatigue, insights
- campaign, roas, cpa, optimize
## Input Contract
Required:
- product_offer
- target_audience
- funnel_stage
- existing_creative_context
Optional:
- brand_guidelines
- prohibited_claims
- placement_constraints
- performance_data
## Output Contract
1. Creative Angle and Script Set
2. Creative Scorecard
3. Fatigue Cycle Assessment
4. Replacement and Test Schedule
5. Inspiration Reference Notes
## Workflow
1. Define creative objective by funnel stage.
2. Generate script options by angle family.
3. Score creatives against performance and compliance criteria.
4. Detect fatigue signals by trend and frequency.
5. Produce replacement and retest sequence.
## Decision Rules
- If CTR drops with rising frequency, mark fatigue risk high.
- If strong hook but weak CVR, revise offer framing before visual overhaul.
- If platform format mismatch exists, adapt script priority by placement.
- If no historical data, start with diversified angle set and faster iteration.
## Platform Notes
Primary scope:
- Meta (Facebook/Instagram), TikTok Ads, YouTube Ads, Google Ads, Amazon Ads
Platform behavior guidance:
- Meta/TikTok: hook speed and first 3 seconds are critical.
- YouTube: longer narrative and proof stacking can work.
- Google/Amazon display: concise value proposition and CTA clarity.
## Constraints And Guardrails
- Avoid unrealistic claims or policy-sensitive copy.
- Separate subjective style feedback from metric-backed conclusions.
- Keep replacement recommendations tied to measurable thresholds.
## Failure Handling And Escalation
- If performance data is absent, output provisional scorecard with confidence label.
- If policy risk is detected, route to Ads Compliance Review.
- If creative production capacity is constrained, prioritize highest-impact variants.
## Code Examples
### Creative Scoring Rubric
weights:
hook_strength: 0.25
offer_clarity: 0.25
visual_attention: 0.20
cta_strength: 0.15
policy_safety: 0.15
### Fatigue Detector
if ctr_drop_pct >= 20 and frequency >= 2.8:
fatigue_level: high
action: rotate_primary_creative
## Examples
### Example 1: New launch script pack
Input:
- Need 5 short-video hooks for cold traffic
Output focus:
- script list
- angle mapping
- CTA variants
### Example 2: Creative score review
Input:
- 15 creatives with last-14-day metrics
Output focus:
- score ranking
- keep/kill suggestions
- next variant ideas
### Example 3: Fatigue replacement timing
Input:
- Top creative slowing down
Output focus:
- fatigue diagnosis
- replacement timing
- backup creative queue
## Quality Checklist
- [ ] Required sections are complete and non-empty
- [ ] Trigger keywords include at least 3 registry terms
- [ ] Input and output contracts are operationally testable
- [ ] Workflow and decision rules are capability-specific
- [ ] Platform references are explicit and concrete
- [ ] At least 3 practical examples are included