skills/albertmartorell1975/android-ai-workflow-foundation/migrate-xml-views-to-jetpack-compose/Gen Agent Trust Hub
migrate-xml-views-to-jetpack-compose
Warn
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [METADATA_POISONING]: The skill metadata explicitly claims the author is "Google LLC", which contradicts the actual author of the skill. This deceptive authorship could mislead users into granting higher trust or permissions than appropriate.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted XML layout files and UI screenshots to generate Kotlin code. This presents an attack surface where instructions embedded in project data could influence the agent's output. \n
- Ingestion points: XML layout files in res/layout and user-provided screenshots or paths (SKILL.md, Step 4). \n
- Boundary markers: None identified; the skill assumes all layout content is valid data. \n
- Capability inventory: Modifying build.gradle, creating Kotlin files, deleting XML and test files, and executing shell commands via testing tools. \n
- Sanitization: No evidence of sanitization or validation for the content of the XML files before processing.
- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The migration workflow involves executing system-level operations such as running Gradle syncs, launching Android emulators, and using testing frameworks like UI Automator and Espresso. These actions represent a significant execution surface on the host machine. Furthermore, the skill suggests using specific versions for Compose and Kotlin (e.g., BOM 2026.06.00, Kotlin 2.3.21) that are significantly ahead of current stable releases and are currently unverifiable.
Audit Metadata