feature-flags

Pass

Audited by Gen Agent Trust Hub on Aug 2, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests data from the project codebase to inform its configuration and generation steps.\n
  • Ingestion points: The skill uses Glob and Grep tools in SKILL.md to scan for existing feature flag, remote configuration, or feature toggle implementations in the project source.\n
  • Boundary markers: There are no boundary markers or delimiters defined to prevent content within these files from being interpreted as instructions by the agent.\n
  • Capability inventory: The skill possesses Write, Edit, and Bash tools, allowing it to modify the project based on the results of its scans.\n
  • Sanitization: The skill does not perform any sanitization or validation of the content discovered in the source files before using it in the generation process.\n- [EXTERNAL_DOWNLOADS]: The generated Swift templates include a RemoteFeatureFlagProvider designed to fetch JSON configuration data from a remote URL endpoint at runtime.\n- [COMMAND_EXECUTION]: The skill utilizes shell discovery tools (Bash, Glob, Grep) to identify the project structure, locate source directories, and verify deployment targets during its activation phase.\n- [SAFE]: The generated Swift code templates implement strong security and concurrency patterns, including @MainActor isolation for UI updates and NSLock for thread-safe access to cached flag values.\n- [SAFE]: Runtime debug overrides and the debug menu view are correctly guarded by #if DEBUG preprocessor directives, ensuring that development-only features do not leak into production builds.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 2, 2026, 01:19 PM
Security Audit — agent-trust-hub — feature-flags