playlet-xhs-feed
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script (
assets/daily_report.py) to process data, generate HTML reports, and automatically open them in the user's default web browser using thewebbrowsermodule. - [EXTERNAL_DOWNLOADS]: The script connects to the RedFox Hub API (
redfox.hk) to retrieve metadata for trending short dramas. This domain is the official infrastructure for the skill's vendor and is used for primary functionality. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-generated content (video titles and descriptions) fetched from Xiaohongshu via an API.
- Ingestion points: External metadata is retrieved in
assets/daily_report.pyand passed to the agent for summarization and trend analysis. - Boundary markers: No explicit delimiter or instruction-bypass warnings are used when presenting external text to the agent.
- Capability inventory: The agent has the ability to run shell commands (via the Python script), write to the local filesystem (
~/Downloadsand~/.workbuddy/cache), and initiate network requests. - Sanitization: While the HTML generation includes standard data formatting, the natural language content is not sanitized against embedded instructions that might influence the agent's summary or analysis.
- [CREDENTIALS_SAFE]: The skill correctly handles authentication by requiring the user to set an environment variable (
REDFOX_API_KEY) instead of encouraging hardcoded secrets.
Audit Metadata