hz-psdk-integration
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes standard local build commands such as
./gradlew assembleDebug,./gradlew lint, and./gradlew testto verify integration health. It also utilizes themetavr(Meta VR CLI) tool for device-related tasks like listing connected headsets, installing APKs, launching applications, and capturing screenshots. These operations are well-scoped to the skill's purpose of Android development for Meta Quest. - [EXTERNAL_DOWNLOADS]: The skill references documentation and SDK artifacts from trusted sources including Meta's developer portal (
developers.meta.com), Meta's API endpoints (graph.oculus.com), and Maven Central (central.sonatype.com). These are official resources required for Horizon Platform SDK integration. - [INDIRECT_PROMPT_INJECTION]: The skill analyzes local codebase files (e.g.,
AndroidManifest.xml,build.gradle, and Kotlin source files) to understand project architecture and dependencies. This creates an attack surface where a malicious project could contain hidden instructions intended to influence the agent's behavior. - Ingestion points: Project files within the path provided by the user during the interactive wizard session.
- Boundary markers: The agent is instructed to cite concrete file paths, but no specific prompt delimiters are used for the content of the files themselves.
- Capability inventory: The skill can read, write, and edit files, and execute shell commands via
gradlewandmetavr. - Sanitization: Not present, as the tool is designed to provide direct implementation and build support for developers.
Audit Metadata