migrate-xml-views-to-jetpack-compose
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to execute shell-level commands within the development environment. Specifically, Step 4 of the workflow instructs the agent to use Android testing frameworks such as UI Automator or Espresso to generate baseline screenshots. Step 5 involves running Gradle sync operations to apply and resolve new Compose dependencies.
- [INDIRECT_PROMPT_INJECTION]: The skill functions by ingesting and analyzing untrusted content from the user's local project, creating a surface for indirect prompt injection.
- Ingestion points: The skill performs a technical audit of all XML layout files within the project's
resdirectories and consumes visual data from screenshots provided by the user or generated by tests. - Boundary markers: The analysis process lacks explicit boundary markers or instructions to treat data within XML files (such as comments or string resources) as non-executable or strictly data-only.
- Capability inventory: The skill possesses significant capabilities, including modifying build configuration files (
build.gradle,libs.versions.toml), executing UI tests, and performing permanent file deletions of XML resources. - Sanitization: The instructions do not specify any validation or sanitization steps for the XML content before it is processed by the agent to generate new Kotlin source code.
Audit Metadata