lev
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is centered around the use of the
levCLI tool, which is designed to execute shell commands, orchestrate workflows, and manage project processes. This functionality is the primary purpose of the skill and aligns with its documentation. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of natural language prompts and structured configurations (such as DNA contracts) that influence agent execution. This creates an attack surface where maliciously crafted data could attempt to influence the agent's behavior during workflow execution.
- Ingestion points: Natural language prompts provided to the
lev execcommand and behavioral constraints defined in DNA contract files. - Boundary markers: The skill recommends using structured prompts with specific sections, but does not enforce formal technical boundaries to isolate untrusted data.
- Capability inventory: The CLI possesses broad capabilities for process execution, file system management, and interaction with local daemons.
- Sanitization: No explicit sanitization or validation for the natural language prompts is described in the skill instructions.
Audit Metadata