optimizing-prompts
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (prompts for optimization), creating a surface for indirect prompt injection. 1. Ingestion points: User-provided text for optimization, as described in SKILL.md (Example 1 and 2). 2. Boundary markers: The skill lacks explicit instructions or markers in SKILL.md or assets/prompt_template.json to distinguish data from instructions. 3. Capability inventory: The skill has broad access to tools listed in SKILL.md including Bash, Read, Write, and Edit. 4. Sanitization: No evidence of input sanitization or validation is present in scripts/README.md or assets.
- [PROMPT_INJECTION]: The file assets/example_prompts.md contains strings mimicking prompt injection attacks, such as 'Ignore previous instructions'. These are documented as examples for detection systems and are not malicious attempts to subvert the agent.
Audit Metadata