The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes project-related metadata and configuration files (e.g., project contracts, studio routing, control models) to generate release checklists. While it ingests external data, the risk is mitigated by its design as a preparation tool that requires manual verification for all critical steps. It explicitly instructs the agent not to perform autonomous actions like uploading resources or enabling production data stores based on ingested data.
[SAFE]: No malicious patterns such as prompt injection, data exfiltration, or obfuscation were detected. The skill maintains a consistent focus on security best practices for game development, including remote authority checks and resource permission audits.

roblox-game-release