The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's core logic in scripts/zoodata.py utilizes subprocess.Popen to manage a background garbage collection worker. This worker is used exclusively for the maintenance task of deleting temporary result files older than 30 days. It re-invokes the script's own path using the current system Python interpreter, which is a secure implementation that does not allow for arbitrary command injection.
[DATA_EXFILTRATION]: The skill transmits operational metadata (ASINs, keywords, and marketplace filters) to the vendor's official API (api.zoodata.ai). The bundled CLI includes built-in safeguards that refuse to transmit the ZOODATA_API_KEY to any untrusted or third-party hosts, ensuring credentials only reach the intended service.
[INDIRECT_PROMPT_INJECTION]: As the skill ingests external data from Amazon listings via the ZooData API, it is technically exposed to indirect prompt injection. However, the skill instructions provide a comprehensive set of boundary markers and output requirements—including confidence labels (📊, 🔍, 💡) and mandatory disclaimers—to ensure the agent treats external data as information rather than instructions.

amazon-daily-market-radar