The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes live search results from Amazon storefronts, representing a surface for untrusted external data to enter the agent's context.
Ingestion points: The scripts/amazon_search.py script retrieves real-time product data (titles, descriptions, etc.) from the Amazon frontend via the vendor's API.
Boundary markers: The instructions do not specify explicit delimiters or warnings to ignore embedded instructions within the search data.
Capability inventory: The skill has the ability to write JSON data files to the project's local directory (linkfox/ subfolder) and perform network requests to vendor APIs.
Sanitization: There is no evidence of sanitization or filtering of the search result content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on shell-based Python scripts to execute its primary functions.
Evidence: SKILL.md and related documentation instruct the agent to run python scripts/amazon_search.py and python scripts/onboarding.py with JSON-formatted arguments.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to multiple subdomains owned by the vendor to facilitate search and authentication.
Evidence: The scripts communicate with tool-gateway.linkfox.com, api.linkfox.com, agent-api.linkfox.com, and skill-api.linkfox.com. These are identified as legitimate vendor resources for 'linkfox-ai'.
[PERSISTENCE]: The onboarding instructions describe methods for persisting environment variables in shell configuration files.
Evidence: references/onboarding.md provides commands to append API keys to ~/.zshrc and ~/.bashrc. While this is a standard configuration practice for CLI tools, it involves modifying shell profiles.

linkfox-amazon-search