macos-spm-app-packaging

Pass

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill's templates include shell scripts that automate developer workflows by interacting with the local system and development tools.
  • compile_and_run.sh and launch.sh use pkill to manage running application instances and open to launch the packaged app.
  • package_app.sh uses standard macOS development tools including swift build, codesign, lipo, and xattr to construct and sign the .app bundle.
  • sign-and-notarize.sh uses xcrun notarytool and stapler to submit the app for Apple's notarization service, which is a trusted process.
  • [INDIRECT_PROMPT_INJECTION]: The skill scaffolds new projects by generating code and configuration files from user-defined metadata.
  • Scripts like package_app.sh generate system files such as Info.plist and .entitlements by interpolating variables like ${APP_NAME} and ${BUNDLE_ID} provided by the user.
  • The SKILL.md instructions guide the agent to perform sed operations on the Package.swift file to rename project targets based on user input.
  • Mitigation: The templates include regex validation (e.g., [[ "$APP_NAME" =~ ^[A-Za-z0-9][A-Za-z0-9_-]*$ ]]) to restrict these variables to safe alphanumeric character sets, preventing command injection or path traversal via project metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 26, 2026, 08:14 PM