prompt-optimizer

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prompts for optimization purposes.
  • Ingestion points: User payloads are ingested in SKILL.md (Step 1) and references/meta-optimization-loop.md.
  • Boundary markers: references/core-patterns.md provides instructions for using XML-style tags and markdown headings to delimit user content.
  • Capability inventory: The skill does not define or use shell commands, network operations, or file-system writing capabilities.
  • Sanitization: The workflow promotes layering and structural separation to mitigate the risk of instruction confusion.
  • [EXTERNAL_DOWNLOADS]: The skill references various official documentation sites and research papers.
  • Evidence: SOURCES.md contains links to OpenAI, Anthropic, Google, and arXiv. These references provide technical context and prompting guidelines.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:03 PM
Security Audit — agent-trust-hub — prompt-optimizer