The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/fetch_rank.py implements a custom HTTPS client function _ssl_post_no_sni that explicitly disables SSL certificate verification and hostname checking (ssl.CERT_NONE and context.check_hostname = False). This practice exposes all communication between the agent and the external API to Man-in-the-Middle (MitM) attacks.
[COMMAND_EXECUTION]: The script scripts/generate_report.py uses subprocess.run with shell=True on Windows systems to automatically open generated HTML files. This pattern can be dangerous if the file path or name is influenced by untrusted external data.
[DATA_EXFILTRATION]: The skill makes network requests to an external, non-whitelisted domain onetotenvip.com via scripts/fetch_rank.py. While only query parameters related to the ranking search are transmitted, the use of a non-standard domain and the bypass of SSL security are notable.
[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection (Category 8) because it processes untrusted data from an external API.
Ingestion points: Data returned from the API call in scripts/fetch_rank.py (e.g., account names, profile URLs).
Boundary markers: None identified in the processing logic or the prompt instructions.
Capability inventory: File system write access, network communication, and the ability to execute shell commands to open files.
Sanitization: scripts/generate_report.py fails to sanitize account names or URLs before interpolating them into the HTML report template, creating a risk of malicious content injection.
[EXTERNAL_DOWNLOADS]: The generated HTML reports reference the html2canvas library hosted on cdn.jsdelivr.net. This is documented as a reference to a well-known and established service.

douyin-rank-tranker