neo-agentic-design

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill consists entirely of Markdown documentation and JSON evaluation files. It provides architectural patterns for LLM orchestration without providing any functional code or implementation scripts.
  • [PROMPT_INJECTION]: The skill does not attempt to override agent behavior. On the contrary, Chapter 18 ('Guardrails & Safety Patterns') in references/advanced-safety.md provides explicit instructions on how to detect and prevent prompt injection and jailbreak attacks.
  • [DATA_EXFILTRATION]: No exfiltration patterns detected. The designs for memory management and inter-agent communication (A2A) are described abstractly using standard protocols like JSON-RPC and HTTP.
  • [NO_CODE]: There are no executable files (e.g., .py, .js, .sh) provided in the skill package. All logic is conceptual and framework-agnostic.
  • [COMMAND_EXECUTION]: There is no use of the dynamic context injection syntax (!command) or any instructions for the agent to execute shell commands. Systems that call tools or execute code are described as sandboxed environments in the reference documents.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 10:43 AM
Security Audit — agent-trust-hub — neo-agentic-design