ios-simulator-skill

Least-privilege permissions

• Run the agent with the minimum user account rights needed—avoid using administrator or root privileges, which would expose your entire computer if compromised.
• Store provisioning profiles (digital certificates that authorize app building) and signing keys in a separate, restricted folder that only the skill can access, not the entire agent system.
• Disable the agent's access to unrelated system folders, especially those containing other projects, passwords, or company data.

Network isolation

• Restrict the agent to communicate only with necessary build systems and app repositories; block outbound connections to unknown or untrusted networks.
• Use a firewall or network sandbox to prevent the skill from sending build artifacts, app binaries, or test data to unexpected destinations.
• Require all communication to use encrypted, authenticated connections (HTTPS/TLS) and validate certificates to prevent interception.

Monitoring

• Log all builds and simulator interactions to a secure audit file—review these logs weekly for unusual commands or outputs you didn't request.
• Set alerts if the skill attempts to access sensitive files like credential stores, email, or payment systems.
• Track which versions of the skill and its underlying libraries (dependencies) are installed; update only after verifying security patches from trusted sources.

Red flags to watch for

• Agent output containing credentials, certificates, or API keys—stop immediately and rotate any exposed secrets.
• Unexpected network traffic or file access to folders outside the project directory.
• Build failures or simulator crashes accompanied by messages asking you to run new commands or visit unfamiliar URLs—this may indicate prompt injection (malicious instructions hidden in app data).
• Dependency updates from unknown sources or changes to the skill without your approval.