adhd-explainer

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is explicitly defined as Markdown-only and does not include any scripts, network operations, or requests for credentials.
  • [PROMPT_INJECTION]: The skill includes robust guidelines for handling untrusted content. For example, the test cases in assets/evals/adhd-explainer-cases.jsonl explicitly require the agent to ignore instructions embedded in pasted text (Indirect Prompt Injection) and refuse requests to reveal its system prompt.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns, package installations, or dynamic code execution (e.g., eval/exec) were detected. Commands mentioned in the documentation are provided as user-facing examples only.
  • [DATA_EXFILTRATION]: There are no network calls or access to sensitive file paths (such as SSH keys or environment variables) within the skill definition.
  • [COMMAND_EXECUTION]: The skill does not perform any shell command execution. The validation.md file mentions development-time testing commands for the author, which do not run in the agent's execution context.
  • [SAFE]: The skill includes a 'Medical boundary' section that strictly forbids the agent from making diagnostic, screening, or treatment claims regarding ADHD, ensuring it remains a purely stylistic tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 07:04 PM
Security Audit — agent-trust-hub — adhd-explainer