The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data, including Amazon product titles, features, and category information retrieved via the vendor's API. This creates a surface for indirect prompt injection where malicious instructions embedded in product metadata could influence the agent's behavior.
Ingestion points: Data returned by the call_api function in scripts/sif_asin_summary.py, specifically fields such as productTitle, productFeatures, and productCategory.
Boundary markers: The skill does not implement explicit boundary markers or instructions to ignore embedded commands within the processed data.
Capability inventory: The skill can write files to the project directory (under the linkfox/ folder), execute Python scripts, and perform network requests to vendor-owned domains.
Sanitization: No explicit sanitization or filtering of the API response content is performed before presentation to the agent.
[PERSISTENCE]: The onboarding documentation provides instructions for users to persist their API credentials by modifying system environment variables and shell configuration files.
Evidence: references/onboarding.md contains instructions for using setx on Windows and appending export commands to ~/.zshrc or ~/.bashrc on macOS and Linux.
[COMMAND_EXECUTION]: The skill relies on executing local Python scripts (scripts/sif_asin_summary.py and scripts/onboarding.py) to perform its core functions, including API communication and account management.
Evidence: SKILL.md defines invocation patterns using python scripts/sif_asin_summary.py and python scripts/onboarding.py.

linkfox-sif-asin-summary