The Agent Skills Directory

[PROMPT_INJECTION] (HIGH): The skill is vulnerable to Indirect Prompt Injection (Category 8) due to its data ingestion patterns and high-privilege capabilities.
Ingestion points: The skill reads untrusted data from testers.csv, testers-internal.txt, and track-config.json via commands like tail and cat.
Boundary markers: Absent. There are no delimiters or instructions to the agent to treat the content of these files as data rather than instructions.
Capability inventory: The skill can execute gplay testers update, gplay release, and gplay edits commit, which allow it to modify Google Play Console tracks, upload application bundles, and manage tester access.
Sanitization: Absent. The skill uses basic shell tools (cut, paste, jq) to format data but performs no validation on the content of the email lists or JSON configurations before pushing them to the Google Play API.
[COMMAND_EXECUTION] (MEDIUM): The skill relies extensively on local shell execution and complex command chaining. It uses shell variables (e.g., $EDIT_ID, $CURRENT, $EMAILS) inside backticks and subshells, which can be risky if the variables are populated with unsanitized data from the external files mentioned above.

gplay-testers-orchestration