system-prompt-creator
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a structured framework for prompt engineering, focusing on task clarity, scope limitation, and fallback handling.
- [SAFE]: The instructions explicitly promote security best practices for generated prompts, requiring that untrusted input variables be isolated with XML delimiters and accompanied by directives to ignore embedded instructions.
- [SAFE]: External references to improvingagents.com in the data format guide are static informational links to retrieval benchmarks and do not involve data exfiltration or automated code execution.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user requirements to generate text deliverables. (1) Ingestion points: SKILL.md workflow mapping requests to input fields. (2) Boundary markers: Absent for the skill's own input ingestion, but mandatory for all generated output prompts. (3) Capability inventory: Limited to text generation; no file-writing, network, or shell execution capabilities. (4) Sanitization: The skill enforces sanitization and isolation rules in its output deliverables, mitigating the risk of downstream injection.
Audit Metadata