The Agent Skills Directory

[SAFE]: The skill's primary function is text generation based on local repository context. It creates a markdown template (production/qa/soak-test-*.md) for human-led testing procedures.
[DATA_EXPOSURE]: The skill reads project-specific documentation such as technical preferences, game design concepts, and previous playtest reports. This behavior is necessary for its stated purpose of generating a relevant testing protocol and does not involve exfiltration to external systems or unauthorized access to system credentials.
[COMMAND_EXECUTION]: While the skill mentions commands like /task-bug or /bug-report, these are documented as internal routing mechanisms or tool aliases for the agent platform to manage project tasks, not for executing shell commands on the underlying host system.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local project files (e.g., playtest reports) to populate the protocol. However, its capabilities are restricted to reading and writing local text files, and it does not possess high-risk capabilities like network access or code execution that could be exploited via injection.

soak-test