intent-guard-shield

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill consists entirely of instructional markdown intended to guide agent behavior toward safer and more transparent task execution. It establishes a workflow for formalizing intent and verifying results before proceeding.
  • [NO_CODE]: No scripts (Python, JavaScript, Shell) or external binary dependencies are included in the skill package. The skill is purely informational/instructional.
  • [PROMPT_INJECTION]: The instructions do not contain patterns to bypass safety filters or override system prompts. Instead, they provide a framework for the agent to explicitly acknowledge constraints and halt execution if they become ambiguous.
  • [DATA_EXFILTRATION]: There are no commands, network operations, or hardcoded credentials that could lead to data exposure or exfiltration. The skill does not access sensitive file paths or environment variables.
  • [REMOTE_CODE_EXECUTION]: No mechanisms for downloading, installing, or executing code from external sources were detected. There are no references to package managers or remote scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 09:20 PM
Security Audit — agent-trust-hub — intent-guard-shield