The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts, including get_pricing.py and get_competitive_pricing.py, utilize subprocess.run to execute a local dependency checker script named check_auth_dependency.py. This execution is restricted to the skill's own file directory and uses the current Python interpreter (sys.executable), which is a legitimate and safe method for verifying internal components.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a data ingestion point, processing responses from the Amazon SP-API which could technically be manipulated by an external entity (an attacker-controlled store). This represents a standard surface for indirect prompt injection.
Ingestion points: External data flows into the skill through the POST /spApi/developerProxy endpoint and command-line arguments provided to scripts.
Boundary markers: The scripts provide raw JSON data or summaries to the agent without explicit boundary delimiters for API content.
Capability inventory: The skill is capable of network operations (API requests to LinkFox domains), local file system writes (saving JSON logs to a linkfox/ directory), and restricted local script execution.
Sanitization: Input parameters such as ASINs and SKUs are properly URL-encoded using urllib.parse.quote and inputs are parsed via JSON, providing protection against basic injection attempts.

linkfox-amazon-store-pricing