The Agent Skills Directory

[SAFE]: The skill is instructional and relies on the Model Context Protocol (MCP) to interact with vendor-provided tools. It does not contain executable code, scripts, or remote downloads.
[PROMPT_INJECTION]: The skill ingests data from external Amazon sources via tools like search_categories and filter_niches. Ingestion points: Data retrieved from Amazon category and niche intelligence tools. Boundary markers: The skill provides instructions for presenting data in structured tables or lists, which reduces the potential impact of malicious content. Capability inventory: The skill has no access to high-privilege capabilities such as arbitrary file writing, subprocess execution, or general network access. Sanitization: No specific sanitization measures are detailed for the ingested data. This surface is characteristic of data-retrieval skills and is considered safe given the agent's restricted toolset.
[CREDENTIALS_UNSAFE]: The skill instructs the agent on how to manage authentication using PANGOLINFO_API_KEY via environment variables or a vendor-specific configuration file. This aligns with standard security practices for secret management in AI tools.

pangolinfo-amazon-niche