gemini-consultation
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION] (HIGH): The documentation promotes the use of
npx -y superconductor-gemini-skills. This command downloads and executes code from an unverified NPM package at runtime. Since the package is not from a trusted organization, this represents an arbitrary code execution risk on the user's system. - [PROMPT_INJECTION] (HIGH): The skill is designed for indirect prompt injection via external file processing.
- Ingestion points: Untrusted PDF and image files provided via the
--fileargument. - Boundary markers: Absent. There are no delimiters or instructions to treat file content as untrusted data.
- Capability inventory: The skill possesses network access (sending data to Google Gemini) and its output influences the agent's context.
- Sanitization: Absent. Content is passed directly to the model.
- [COMMAND_EXECUTION] (MEDIUM): The skill executes shell scripts and npx commands. If the input "YOUR QUESTION" or the
--filepath is not strictly sanitized by the underlying script, it could allow for local command injection.
Recommendations
- AI detected serious security threats
Audit Metadata