ito-data-atlas-agent
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of markdown documentation and architectural patterns. No executable scripts or code are provided.
- [NO_CODE]: The skill does not contain any Python, JavaScript, or shell scripts, limiting the potential for direct technical exploits within the skill itself.
- [INDIRECT_PROMPT_INJECTION]: The architecture describes an agent that ingests data from external sources such as the web, X, and GitHub. While this introduces an attack surface for the resulting agent, the skill explicitly includes a 'Risk reviewer' role to mitigate prompt injection risks and requires human-in-the-loop approvals.
- [CREDENTIALS_UNSAFE]: Mentions the requirement for an
ITO_API_KEYin the resulting agent's configuration, but emphasizes using it for read-only access and cautions against exposing credentials in public docs. No actual secrets are hardcoded.
Audit Metadata