The Agent Skills Directory

[DATA_EXPOSURE]: The script scripts/check_sensitive_words.py attempts to read sensitive shell profile files including ~/.zshrc, ~/.bashrc, ~/.bash_profile, and ~/.profile to extract the REDFOX_API_KEY. Accessing these files can expose other environment variables or sensitive configurations stored by the user.
[COMMAND_EXECUTION]: The skill requires the execution of shell commands for installation and operation, specifically playwright install chromium to enable web page content extraction.
[DATA_EXFILTRATION]: The skill sends user-provided text content to an external API at https://redfox.hk/story/api/cozeSkill/sensitiveWordSearch for analysis. It also performs outbound network requests to arbitrary URLs provided by the user to extract text for compliance checking.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from uploaded files (DOCX, TXT), images, and external websites. It lacks explicit instructions to ignore potentially malicious commands embedded within the analyzed text.
Ingestion points: Text content extracted from user-uploaded files, external URLs via Playwright, and image text extraction.
Boundary markers: Absent. The skill does not use delimiters or instructions to prevent the agent from following commands found within the ingested text.
Capability inventory: The skill can perform network requests (via requests and playwright) and write new text files to the local file system (e.g., ./抖音_优化文案_*.txt).
Sanitization: The script performs basic HTML tag removal but does not sanitize the logical content of the text before it is processed by the agent or written to output files.

douyin-prohibited-word