braintrust-red-team-agent

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is primarily instructional and establishes a security-conscious framework for red teaming. It mandates authorization checks and scoped testing to prevent unauthorized or destructive actions.
  • [PROMPT_INJECTION]: A static detector flagged a potential injection pattern in references/interaction-contract.md. However, manual analysis reveals this is a defensive instruction (Rule 9) that explicitly warns the agent against being influenced by 'ignore previous instructions' commands found in analyzed data traces. This is a recommended safety practice.
  • [NO_CODE]: The skill does not contain executable code, script files, or remote dependencies, significantly reducing the attack surface for remote code execution or persistence.
  • [DATA_EXFILTRATION]: The skill does not perform network operations or access sensitive system paths. It provides templates for findings that recommend storing evidence in controlled, non-inline locations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 10:16 PM