The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted third-party content (Amazon product reviews, features, and descriptions) retrieved via the ZooData API.
Ingestion points: Data enters the agent context through the realtime/product and reviews/analysis endpoints in scripts/zoodata.py.
Boundary markers: The workflow templates in references/scenarios-listing.md and references/scenarios-eval.md do not implement explicit delimiters or 'ignore embedded instructions' warnings when processing this external content.
Capability inventory: The skill can execute various subcommands and shell operations through the bundled scripts/zoodata.py CLI utility.
Sanitization: The Python transport script fetches and formats JSON data from the API but does not perform explicit sanitization or escaping of the text content before it is presented to the agent.
[COMMAND_EXECUTION]: The script scripts/zoodata.py uses subprocess.Popen to perform an internal background maintenance task.
Evidence: In _schedule_stale_result_cleanup, the script spawns a detached process to invoke itself with a specific cleanup flag (RESULT_CLEANUP_WORKER_FLAG) using fixed arguments.
Security Context: This self-invocation is used for routine cleanup of temporary result directories and does not accept user-controlled input, effectively mitigating command injection risks.

amazon-analysis