skills/spillwavesolutions/mastering-gemini-cli-agentic-skill/mastering-gemini-cli/Gen Agent Trust Hub
mastering-gemini-cli
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The documentation provides technical details on the RunShell tool and the --approval-mode yolo flag, which allows for autonomous shell command execution. The skill explicitly warns users about the potential risks (such as destructive commands) and strongly recommends using containers or sandboxed environments for such workflows.
- [INDIRECT_PROMPT_INJECTION]: The skill documents how to build agents that ingest project files, which creates a surface for indirect prompt injection. It addresses this by instructing users on how to define constraints in GEMINI.md and how to limit agent capabilities using the --allowed-tools whitelist and human-in-the-loop approval settings. 1. Ingestion points: Files accessed via ReadFile or directory inclusion. 2. Boundary markers: System instructions and constraints defined in GEMINI.md. 3. Capability inventory: File reading, writing, editing, and shell execution capabilities. 4. Sanitization: Instructions emphasize capability restriction via the --allowed-tools flag.
- [DYNAMIC_EXECUTION]: The assets/headless-wrapper.sh template uses eval to run a shell command constructed from user prompts. While a standard utility pattern for local scripts, it is documented as a template for user adoption and is not an automated payload.
Audit Metadata