mobile-app

Fail

Audited by Gen Agent Trust Hub on Oct 8, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to install core dependencies using insecure piped shell execution patterns, such as curl -fsSL https://asccli.sh/install | bash for the asc CLI and curl -fsSL https://raw.githubusercontent.com/tamtom/play-console-cli/main/install.sh | bash for the gplay CLI. It also includes an Invoke-Expression (iex) pattern for Windows PowerShell installation of these tools.
  • [EXTERNAL_DOWNLOADS]: The orchestration logic depends on a wide array of sub-skills and tools from third-party GitHub repositories not recognized as trusted vendors, including rorkai, tamtom, AvdLee, Dimillian, twostraws, callstackincubator, and expo. The skill also suggests cloning and building the rork-local tool from source if its npm package is missing, which increases supply chain risk.
  • [CREDENTIALS_UNSAFE]: The skill handles highly sensitive authentication artifacts like App Store Connect API keys (.p8), Google Play service account JSON files, and keystores. While it advises against committing these to the repository, the agent is tasked with managing their file paths and interacting with them via CLI commands.
  • [COMMAND_EXECUTION]: The skill frequently invokes shell commands through the asc and gplay tools to perform operations such as registering bundle IDs, creating app records, and promoting releases to production tracks.
  • [INDIRECT_PROMPT_INJECTION]: The intake process ingests free-form user descriptions of the app's functionality to drive subsequent code generation and store metadata setup.
  • Ingestion points: Application description intake in SKILL.md (Stage 0).
  • Boundary markers: Absent.
  • Capability inventory: Includes shell command execution, tool installation (npx, npm), and filesystem writes.
  • Sanitization: No sanitization or validation of the user-provided input is specified before it is used to drive the orchestration of sub-skills.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Oct 8, 2026, 11:02 PM