deploy-mobile-apps

Warn

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The scripts mobile_release_lib.py and mobile_release.py extensively use subprocess.run to invoke system tools including git, gh (GitHub CLI), and flutter. This provides the skill with broad execution capabilities within the local environment to manage worktrees and dependencies.
  • [PRIVILEGE_ESCALATION]: The skill implements a guarded merge mechanism in mobile_release_lib.py that utilizes gh pr merge --admin. This allows the agent to perform administrative merges that bypass standard repository branch protection rules and mandatory pull request reviews.
  • [CREDENTIALS_UNSAFE]: The skill is designed to manage and utilize sensitive API tokens. scripts/codemagic.py includes logic in the _keychain_token function to programmatically retrieve the CODEMAGIC_API_TOKEN from the macOS login keychain using the security find-generic-password command. It also accesses this token through environment variables.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external configuration data, creating a potential surface for indirect injection via manipulated repository files.
  • Ingestion points: The skill reads codemagic.yaml, release-please-config.json, and pubspec.lock directly from the managed application repositories using git show.
  • Boundary markers: The skill does not employ explicit boundary markers or instructions to the agent to ignore potentially malicious content embedded in these configuration files.
  • Capability inventory: The skill has the ability to write to the file system, perform git push/merge operations, and interact with the CodeMagic API.
  • Sanitization: While the skill parses data using structured formats (JSON/YAML), it lacks specific validation to prevent logical manipulation of the deployment workflow by an attacker with commit access to the development branches.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 22, 2026, 07:21 PM