自动抓取并分类多平台每日游戏资讯，生成格式化报告发送飞书并存档为Word和文本文件。

What to check before installing or running this skill: - Inspect the config file at /home/admin/.openclaw/workspace/configs/news-crawler-config.json before running: the code reads that file and it may contain URLs or credentials. Do not run the skill until you understand what is in that file. - Feishu push mismatch: SKILL.md/README say reports are sent to Feishu, but the package does not declare any Feishu tokens or webhook env vars. Determine where you must put your Feishu credentials (likely in the config file) and ensure they are safe. - Network and subprocess risk: the scripts call curl, call external services (r.jina.ai/jina.ai), run 'uv run' and run other local scripts (searxng path). If you cannot trust those external endpoints, run the skill in an isolated/test environment first. - Cron change is manual: the README suggests installing a cron job. Only add the cron entry if you trust the code and have reviewed the scheduled command. - Review sending code: search the codebase for any HTTP POST/requests that transmit data (especially to external endpoints) and verify exactly what is sent (reports, system paths, config contents). Confirm no sensitive tokens are transmitted inadvertently. - Run manually first: execute the crawler in a sandbox or VM and observe outbound connections, files created under /home/admin/.openclaw/workspace/reports and logs, and check for failures that may indicate missing dependencies or expected local services (SearXNG). - If you need this skill but want least privilege: remove or blank any push/webhook config until you explicitly provide a Feishu webhook, and avoid installing the cron until after testing.

Type: OpenClaw Skill Name: daily-game-news Version: 2.0.0 The skill bundle is a comprehensive game news aggregator designed to scrape multiple websites (e.g., GCORES, IGN, GameSpot) and generate reports in Markdown, Word, and Text formats. It utilizes various fetching strategies including RSS, direct HTML parsing with BeautifulSoup, and external tools like SearXNG and Jina Reader (r.jina.ai). While the scripts frequently use `subprocess.run` to execute `curl` and internal helper scripts, they do so using safe argument lists rather than shell strings, and the targets are limited to the configured news sites. No evidence of data exfiltration, credential theft, or malicious prompt injection was found; the skill's behavior aligns entirely with its stated purpose of automated news reporting.