capture-skill
Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill facilitates the creation of new prompt files based on untrusted external input (user-pasted prompts or descriptions). \n- Ingestion points: Source material provided by the user in Step 1 (SKILL.md). \n- Boundary markers: No specific delimiters or warnings are used when processing input to distinguish data from instructions. \n- Capability inventory: The skill performs a file write operation to .kiro/prompts/{name}.md (SKILL.md). \n- Sanitization: Validates the file name but does not sanitize the content of the generated prompt for malicious instructions.
Audit Metadata