koray
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external content which may contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: The "Before Starting" section in
SKILL.mdinstructs the agent to gather "The article — URL or content to optimize" and "Top 3-5 competitor URLs". - Boundary markers: No specific delimiters or instructions to ignore embedded commands within the processed content are provided in
SKILL.md. - Capability inventory: The skill suggests fetching content via "REST API or SSH", providing the agent with powerful capabilities to interact with remote infrastructure.
- Sanitization: The instructions lack any requirement for sanitizing, escaping, or validating the content retrieved from external URLs or via SSH before the agent incorporates it into its context.
- [COMMAND_EXECUTION]: The skill explicitly encourages the agent to use high-privilege protocols that can be leveraged for broader system access.
- Evidence:
SKILL.mdcontains the instruction: "If the user provides a WordPress post URL on your site, you can fetch the content via REST API or SSH." The use of SSH for content retrieval is a high-privilege operation that could be misused for unauthorized command execution if the agent is not properly constrained.
Audit Metadata