The Agent Skills Directory

[SAFE]: The skill makes network requests to redfox.hk for data retrieval. This domain is the official endpoint for the RedFox Hub service provided by the skill's author, redfox-data, and represents expected vendor functionality for an API-driven skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from various social media platforms (Xiaohongshu, Douyin, WeChat) and AI search results (Doubao, Kimi, DeepSeek), creating an indirect prompt injection surface. The risk is mitigated by the following measures:
Ingestion Points: Content is fetched via API calls in scripts/cn30_search.py, scripts/hub_fetch.py, and scripts/geo_search.py and stored in local JSON files.
Boundary Markers: During the AI-assisted analysis phase, the skill uses structured prompt templates in scripts/lib/analyzer.py that enforce a strict JSON schema for the agent's output, helping to separate data from instructions.
Capability Inventory: The skill is authorized to use Bash, Read, Write, AskUserQuestion, and WebSearch tools, allowing it to generate local HTML reports and perform supplemental research.
Sanitization: The skill includes robust sanitization routines; for example, scripts/lib/platforms.py strips internal engine citation tags (like <REF>), and scripts/cn30_search.py performs HTML escaping for generated reports.
[EXTERNAL_DOWNLOADS]: The skill identifies a dependency on the requests library in SKILL.md. This is a standard, trusted package from the official Python Package Index (PyPI) and is used appropriately for API communication.

brand-geo-analysis-plus