agent-creator
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches the official sub-agent specification from docs.openhands.dev to ensure it uses the most up-to-date schema and path conventions.
- [COMMAND_EXECUTION]: The skill uses the file_editor tool to save generated agent files to the project-level directory (.agents/agents/) or the user's home directory (~/.agents/agents/) as appropriate for the selected scope.
- [INDIRECT_PROMPT_INJECTION]: As a generator skill, it accepts user-provided text and incorporates it into a new agent system prompt, which represents a potential injection surface if the requirements are malicious.
- Ingestion points: User-provided responses to the interview questions in SKILL.md.
- Boundary markers: Absent; responses are interpolated directly into the generated markdown template.
- Capability inventory: The skill has the capability to write files to specific agent directories.
- Sanitization: The skill does not perform sanitization on user input before embedding it into the generated agent prompt.
Audit Metadata