The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has an inherent attack surface because it is designed to ingest and process untrusted data from local project files (source code, manifests, and configuration files) using tools like Read and Grep. While this could allow maliciously crafted project files to influence the agent's behavior, it is the primary intended function of the skill. No specific boundary markers or sanitization steps are defined to mitigate this.
[COMMAND_EXECUTION]: The skill uses the Bash tool and provides instructions to execute commands for project discovery and analysis. These commands are limited to local file system operations necessary for the audit and do not involve network exfiltration or unauthorized privilege escalation.
[DATA_EXFILTRATION]: The skill searches for sensitive files (e.g., .env, google-services.json, keystores) and hardcoded secrets as part of its security hygiene audit. However, there are no instructions or network patterns that suggest this information is sent to external servers.
[SAFE]: The skill follows professional security auditing practices by instructing the agent to verify privacy policies, data safety declarations, and account deletion features.

playstore-review