show-me
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-controlled data via the
topicparameter to generate visual representations, creating a surface for indirect prompt injection. - Ingestion points: The
topicparameter defined in bothSKILL.mdandREADME.mdserves as the entry point for untrusted data. - Boundary markers: Absent. The instructions do not direct the agent to use delimiters or ignore any instructions that might be embedded within the provided topic.
- Capability inventory: The skill instructions involve executing shell commands to launch applications and using automation tools like Playwright to drive UI states.
- Sanitization: Absent. There are no requirements or steps provided to sanitize, escape, or validate the content of the topic before it is processed by the agent.
- [COMMAND_EXECUTION]: The screenshot functionality requires the agent to execute commands to run the target application and utilize Playwright or browser extensions for automation. While this is an intended feature, it utilizes the system's command execution capabilities to interact with local processes and browser environments.
Audit Metadata