debug-generated-project
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- REMOTE_CODE_EXECUTION (HIGH): The skill clones the 'tuist/tuist' repository and performs a 'swift build' to create a binary which it then executes on the host. Since the 'tuist' organization is not in the trusted sources list, this is considered execution of untrusted remote code.
- EXTERNAL_DOWNLOADS (HIGH): The skill utilizes 'mise' to download and execute the latest Tuist binary and 'git' to download source code from an external repository not on the trusted list.
- COMMAND_EXECUTION (MEDIUM): The skill executes various system-level commands including 'xcodebuild', 'zip', and 'simctl' (to boot and launch applications) based on the context of the project being debugged.
- PROMPT_INJECTION (LOW): (Category 8: Indirect Prompt Injection) The skill processes user-provided 'Project.swift' and 'Tuist.swift' files. Ingestion points: User manifest content. Boundary markers: None present in scripts. Capability inventory: 'tuist generate' (evaluates manifests), 'xcodebuild' (compiles code), 'simctl launch' (runs apps). Sanitization: None performed on user-provided Swift code before execution.
Recommendations
- AI detected serious security threats
Audit Metadata