prompt-engineering

Pass

Audited by Gen Agent Trust Hub on Aug 1, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is entirely educational and descriptive, focusing on best practices for prompt engineering such as persona definition, few-shot prompting, and JSON structuring.
  • [EXTERNAL_DOWNLOADS]: Includes references to Anthropic's official API endpoint (api.anthropic.com) for documentation purposes. This is a well-known service reference and is considered safe.
  • [INDIRECT_PROMPT_INJECTION]: The skill describes a workflow that processes external inputs (e.g., article titles from a webhook). While this creates an input surface, it is a standard implementation for automation tasks and the skill does not exhibit malicious behavior.
  • Ingestion points: External data (article titles) entering the agent via Make.com webhook triggers (SKILL.md).
  • Boundary markers: Uses labels like 'Título para o Artigo:' to delineate input, though it lacks explicit instructions to ignore embedded commands within the input variable.
  • Capability inventory: The skill documentation mentions external network operations (HTTP POST) via Make.com but does not execute them itself.
  • Sanitization: None specified in the provided prompt templates.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 1, 2026, 12:01 PM
Security Audit — agent-trust-hub — prompt-engineering