Agent Skills › leamsigc/ShortsGenerator › reddit-research-scrape

reddit-research-scrape

GitHub

从Reddit抓取帖子数据并生成病毒式视频脚本。通过JSON API获取准确指标，调用后端API生成西班牙语脚本，输出结构化Markdown和索引文件。

skills/reddit-research-scrape/SKILL.md leamsigc/ShortsGenerator

Trigger Scenarios

需要分析Reddit帖子热度为Reddit内容生成短视频脚本批量采集Reddit社区数据

Install

npx skills add leamsigc/ShortsGenerator --skill reddit-research-scrape -g -y

More Options

Use without installing

npx skills use leamsigc/ShortsGenerator@reddit-research-scrape

指定 Agent (Claude Code)

npx skills add leamsigc/ShortsGenerator --skill reddit-research-scrape -a claude-code -g -y

安装 repo 全部 skill

npx skills add leamsigc/ShortsGenerator --all -g -y

预览 repo 内 skill

npx skills add leamsigc/ShortsGenerator --list

SKILL.md

Frontmatter

{
    "name": "reddit-research-scrape",
    "version": "3.0.0",
    "description": "Scrape state — runs the Reddit JSON API scraper for correct data (upvotes,\ncomment counts, full comment trees, exact selftext), then generates a viral\nvideo script per post from the post content + top comments via the backend\nAPI. Saves structured markdown files + index.json.\n",
    "allowed-tools": [
        "Bash",
        "Read",
        "Write",
        "ChromeDevTools"
    ]
}

Reddit Research — Scrape State (v3)

Scrapes top posts from a subreddit with correct upvotes/comment counts using Reddit's JSON API, then generates a viral video script per post.

Before running

Ensure the Flask backend is running on port 8080 (for script generation): curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/api/models

Run

node "$CLAUDE_SKILL_ROOT/scripts/scrape.js" <subreddit> [limit] [sort] [lang] [--session <id>] [--data-root <path>]
# fallback: node ./skills/reddit-research/scripts/scrape.js <subreddit> ...

Defaults: limit=10, sort=hot, lang=es (Spanish).

What the script does

Fetches the subreddit listing JSON: https://www.reddit.com/r/<sub>/<sort>.json?limit=<n>
For each post, fetches <permalink>.json for correct data:
- Accurate upvotes (not the HTML bug showing 1)
- Accurate comment counts, full comment trees with correct scores
- Exact selftext, video URL (media.reddit_video.fallback_url), linked URL
Extracts the top 10 comments (depth ≤ 2)
Generates a viral video script via the backend API (title + content + top comments), ~30-60s
Writes .reddit-research/<YYYY-MM-DD_HHMMSS>/<post-slug>.md (dated session) with:
- # Title, author, upvotes, comments, source, scraped date
- ## Content, ## Viral Video Script (Spanish), ## Videos, ## Sourced Videos (placeholder), ## Top Comments
Writes index.json (posts[] with index/title/file/url/upvotes/comments)

Manual fallback (Chrome DevTools CLI)

If the JSON API is blocked, scrape the listing page manually:

chrome-devtools navigate_page --url "https://www.reddit.com/r/<subreddit>/<sort>/" — scroll to load more posts
Extract post URLs from a[href*="/comments/"]
Fetch each post's data via fetch("<post_url>.json") in evaluate_script
Write the same markdown + index.json structure