The Agent Skills Directory

[DATA_EXFILTRATION]: The script scripts/fetch_explosive_articles.py performs network operations to the domain onetotenvip.com. It uses raw Python socket and ssl modules to manually construct HTTPS requests instead of the standard requests library. This implementation explicitly disables SSL certificate verification by setting verify_mode to ssl.CERT_NONE, which exposes the connection to Man-in-the-Middle (MitM) attacks.
[COMMAND_EXECUTION]: The script scripts/fetch_explosive_articles.py uses subprocess.run to execute a local Python script scripts/generate_html.py. It passes potentially untrusted data fetched from the external API as arguments to this subprocess.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. It retrieves data from an external third-party API and is instructed by core_workflow.md to output the resulting content verbatim to the user. This instruction could lead the agent to follow malicious commands hidden in the fetched data.
Ingestion points: Data fetched from onetotenvip.com in scripts/fetch_explosive_articles.py.
Boundary markers: Absent. Instructions specifically command the agent to output external content without modification.
Capability inventory: Subprocess execution in scripts/fetch_explosive_articles.py and file generation.
Sanitization: Only basic whitespace and character stripping is performed; there is no validation against malicious instructions.
[EXTERNAL_DOWNLOADS]: The assets/preview-template.html file includes external JavaScript dependencies (html2pdf.js) from various CDN providers including jsdelivr.net, cloudflare.com, and unpkg.com.

xhs-breaking-rankings