spec-to-code-compliance

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is inherently vulnerable to indirect prompt injection (Category 8) due to its core function of processing untrusted external data sources.
  • Ingestion points: The skill explicitly instructs the agent to discover and normalize external documents (PDF, Markdown, DOCX, HTML) and codebase files in Phase 0 and Phase 1.
  • Boundary markers: There are no specific instructions or delimiters provided to the agent to treat instructions found within the audited documentation as data rather than instructions (e.g., a 'whitepaper' could contain a hidden instruction to the auditor to ignore certain files).
  • Capability inventory: The skill is designed for high-precision semantic analysis and reporting. While it does not explicitly invoke high-risk shell or network tools in the provided text, the platform capabilities available to the 'spec-compliance-checker' agent could be exploited if it obeys instructions inside the spec documents.
  • Sanitization: No sanitization, escaping, or validation steps for the content of processed documents are defined before the content enters the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 08:58 AM