play-billing-library-version-upgrade
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to processing untrusted project files while holding build execution capabilities.
- Ingestion points: The skill ingests untrusted project data by reading build configurations (
build.gradle,build.gradle.kts,libs.versions.toml) and scanning application source files during the version discovery and refactoring stages inSKILL.md. - Boundary markers: There are no explicit boundaries or isolations defining where the untrusted file content ends and the core instructions begin, nor are there instructions to ignore natural language commands found within the source code.
- Capability inventory: The skill maintains capabilities to execute local development commands including
./gradlew assembleDebug,./gradlew test,./gradlew clean assembleDebug,./gradlew sync, and./gradlew buildas specified inSKILL.md(Phase 3). - Sanitization: There is no specified logic for sanitizing or filtering string inputs extracted from the target application before interpreting project artifacts.
Audit Metadata