prompt-enhancer

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is intended to intake raw, untrusted user prompts to perform enhancement and optimization, making it susceptible to instructions embedded in that input.
  • Ingestion points: The skill takes a user-supplied prompt as its primary input for transformation (e.g., refactoring, compression, or adding constraints).
  • Boundary markers: The instructions advocate for wrapping structured data in XML tags (like <constraints> and <context>), which serves as a basic delimiter, though the main mission remains in natural language.
  • Capability inventory: The skill body focuses on text transformation and does not specify access to external tools, network operations, or file-system writing capabilities in its frontmatter.
  • Sanitization: The skill lacks explicit guidance for the agent to detect, ignore, or sanitize malicious instructions that might be contained within the user's input text during the enhancement process.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 05:05 PM
Security Audit — agent-trust-hub — prompt-enhancer