gradle-run
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The scripts/gradle_run.py script executes Gradle commands using subprocess.Popen. The implementation avoids shell=True and validates that the command starts with an approved Gradle launcher (e.g., gradle, ./gradlew).
- [DYNAMIC_EXECUTION]: The test script scripts/test_gradle_run.py uses importlib to dynamically load the local scripts/gradle_run.py for testing. This is a standard testing pattern for local files.
- [INDIRECT_PROMPT_INJECTION]: The skill processes output from external Gradle builds, creating a surface for potential indirect prompt injection. * Ingestion points: scripts/gradle_run.py reads Gradle output logs in extract_diagnostics. * Boundary markers: SKILL.md instructs the agent to use only the bounded JSON summary output. * Capability inventory: The skill executes subprocesses in scripts/gradle_run.py. * Sanitization: scripts/gradle_run.py implements a redact function to mask credentials (passwords, tokens, authorization headers) in build output before summarization.
Audit Metadata