The Agent Skills Directory

[Command Execution]: The skill utilizes git rev-parse --show-toplevel as a fallback mechanism to identify the repository root. This is a standard utility command for environment discovery and does not pose a security risk in this context.
[Indirect Prompt Injection Surface]: The instructions direct agents to read shared files like decisions.md and history.md. Since these files are populated by other agents or external processes, they represent an indirect prompt injection surface where instructions from one agent could potentially influence another. This is a characteristic of collaborative workflows rather than a direct vulnerability, but it is a consideration for data integrity.
[File System Operations]: The skill defines patterns for writing decision files to specific paths within the .squad/ directory. These operations are scoped to the project's internal workflow and do not involve sensitive system paths.

agent-collaboration