The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides explicit instructions to download and execute code from an untrusted GitHub repository (https://github.com/NanmiCoder/MediaCrawler.git). This involves cloning the project and installing its dependencies using pip and uv, allowing for the execution of unverified third-party scripts.
[COMMAND_EXECUTION]: The skill requires the agent to execute several shell commands for repository management, environment creation, and browser automation setup, such as playwright install chromium.
[EXTERNAL_DOWNLOADS]: The skill downloads the MediaCrawler repository and third-party Python packages from unverified sources.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from various external social media platforms (Xiaohongshu, Douyin, Zhihu, etc.). This untrusted data represents an attack surface for indirect prompt injection, as the agent may process malicious instructions embedded in the scraped content. \n
Ingestion points: Data scraped from social media platforms mentioned in SKILL.md and references/media-crawler-deep-dive-20260517.md. \n
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are provided for the scraped data. \n
Capability inventory: The skill utilizes terminal for repository cloning and dependency installation, browser for platform automation, and network operations via proxies or CDP for scraping. \n
Sanitization: The provided files do not include mechanisms for sanitizing or validating the content retrieved from external platforms.
[DYNAMIC_EXECUTION]: The skill utilizes browser automation via Playwright and the Chrome DevTools Protocol (CDP) to interact with web pages and execute JavaScript to bypass anti-crawling signatures.

social-media-crawler