Agent Skills
› leamsigc/ShortsGenerator
› ig-research-report
ig-research-report
GitHub基于抓取数据生成美观的HTML社交媒体研究报告，包含项目概览、互动统计、热门帖子卡片及视觉钩子分析。自动检测评论引导等成功模式，支持任意平台原始数据，输出深色主题报告。
Trigger Scenarios
需要生成社交媒体内容分析报告
将抓取的数据转换为可视化HTML报告
分析视频帖子的视觉钩子和成功模式
Install
npx skills add leamsigc/ShortsGenerator --skill ig-research-report -g -y
SKILL.md
Frontmatter
{
"name": "ig-research-report",
"version": "1.0.0",
"description": "Report state — generates a beautiful HTML report from scraped data.\nPlatform-agnostic: works with raw-posts.json from any platform.\nReads transcripts, hook screenshots, engagement data, and produces\na dark-themed report with post cards, pattern analysis, and findings.\n",
"allowed-tools": [
"Bash",
"Read"
]
}
Social Research — Report State
Generates a styled HTML report from scraped data, transcripts, and screenshots.
Run
node "$CLAUDE_SKILL_ROOT/scripts/report-html.js" <project-name> [sessionId] # resolves latest if no session
What it generates
The report includes:
- Hero section: Project name, niche, search terms, scan date
- Stats bar: Total posts, reel %, transcribed count, visual hooks, top likes
- Top 6 post cards: Rank, author, engagement, hook screenshots (0s/1s/2s), visual hook description, spoken hook quote, "why it worked" analysis, full transcript (scrollable), caption preview
- Winning patterns: Format breakdown, comment CTA usage, hook analysis, content themes
"Why it worked" detection
Automatically checks for:
- Comment-bait CTAs
- Specific numbers in hooks (anchoring)
- Named frameworks (authority)
- Short punchy hooks (≤8 words)
- Question format (curiosity gap)
- Comparison format (contrast)
- Time promises (loss aversion)
Visual hook inference
Infers from caption/transcript: split screen, whiteboard, screen recording, talking head, before/after, etc.
Output
.ig-research/projects/<project-name>/report.html
Version History
- 7b83166 Current 2026-08-29 02:30
