skill-creator
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes npx @nextstage-brasil/harness sync to update project-level skill symlinks. This is a legitimate utility provided by the skill's author for managing agent environments.\n- [EXTERNAL_DOWNLOADS]: It references and incorporates logic from the anthropics/skills repository to provide core skill-creation functionality, such as evaluation loops and description optimization. These are established resources from a trusted organization.\n- [PROMPT_INJECTION]: The skill acts as a factory for generating new agent skills, which creates a surface for indirect prompt injection where user-supplied instructions could lead to the generation of skills with unsafe behaviors.\n
- Ingestion points: User prompts and descriptions for new skills provided during the creation or improvement workflow.\n
- Boundary markers: The manifest does not define explicit delimiters to isolate user-provided data from the generated instructions.\n
- Capability inventory: File system write access to the .agents/skills/ directory and the ability to execute shell commands via npx.\n
- Sanitization: The wrapper relies on the agent's internal safety protocols and the logic of the upstream anthropics/skills framework to validate generated content.
Audit Metadata