The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell scripts and subprocess calls to manage its environment.\n
scripts/check-cdp.sh manages Chrome processes and starts Chromium with remote debugging.\n
scripts/check-agent-browser.sh executes system commands to install Node.js and global NPM packages.\n
scripts/crawl_xueqiu_timeline_api.py uses subprocess.run to execute the agent-browser CLI tool.\n- [EXTERNAL_DOWNLOADS]: The skill downloads and installs software during setup.\n
scripts/check-agent-browser.sh installs node@22 via Homebrew.\n
scripts/check-agent-browser.sh fetches and installs the agent-browser package globally from the well-known mirror https://registry.npmmirror.com.\n- [REMOTE_CODE_EXECUTION]: The skill performs dynamic code execution within a browser context.\n
scripts/crawl_xueqiu_timeline_api.py utilizes agent-browser eval to run dynamically constructed JavaScript snippets in the browser to interact with Xueqiu's API. The user_id variable is validated via regular expressions before being interpolated into the script.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests untrusted data that could be analyzed by an AI agent later.\n
Ingestion points: Xueqiu User Timeline API (xueqiu.com/v4/statuses/user_timeline.json) via the get_api_data_in_browser function in scripts/crawl_xueqiu_timeline_api.py.\n
Boundary markers: No specific delimiters or safety instructions are used when saving the crawled content to Markdown files.\n
Capability inventory: The skill can perform file system writes and execute network-connected browser automation commands.\n
Sanitization: The clean_html function removes HTML tags but does not filter for malicious linguistic patterns or instructions in the text content.

crawl-xueqiu-timeline