prompt-template-builder

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides structural templates and best practices for prompt engineering. No malicious code, obfuscation, persistence mechanisms, or unauthorized network operations were found in the instructions or code snippets.
  • [INDIRECT_PROMPT_INJECTION]: The templates are designed to interpolate untrusted data (such as user-submitted code) into LLM prompts, which is a common attack surface for indirect prompt injection.
  • Ingestion points: Untrusted data enters the context via the variables.code and variables.context fields in the template construction logic within SKILL.md.
  • Boundary markers: The provided templates use Markdown code blocks (triple backticks) to delimit external content, which provides a basic boundary but does not fully prevent instruction override.
  • Capability inventory: The skill contains no implementation of network requests, file system modifications, or shell command executions; its capabilities are restricted to string manipulation and data validation.
  • Sanitization: The logic does not perform sanitization on input variables before interpolation, although it defines a strict Zod schema to validate the structure of the resulting LLM output.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:14 PM
Security Audit — agent-trust-hub — prompt-template-builder