slopship
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill performs command execution for building and verifying code via the
slopshipper verify --cmdinterface. This functionality is the primary purpose of the implementation workflow. - [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill requires a pre-installed
slopshipperbinary and contains explicit instructions forbidding the agent from downloading or installing software, ensuring the environment remains under user control. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill interacts with external services like GitHub to deliver pull requests as part of the artifact delivery process. It mitigates local data exposure by using stdin for evidence payloads instead of writing them to the file system.
- [INDIRECT_PROMPT_INJECTION]: The skill processes implementation plans and CLI status documents. Ingestion points: User-provided plans and
slopshipper statusJSON. Boundary markers: Structured JSON schemas and human 'release' gates. Capability inventory: Shell command execution viaslopshipper verify. Sanitization: Forced dry-run validation and human oversight requirements.
Audit Metadata