braintrust-report-eval-results

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists entirely of documentation and instructional guidelines in Markdown format. There are no executable scripts, network operations, or external dependencies.
  • [PROMPT_INJECTION]: While the interaction contract mentions phrases like 'ignore previous instructions', these are part of a defensive policy instructing the agent to ignore such content when found in data traces. This is an intended security measure, not an attack.
  • [INDIRECT_PROMPT_INJECTION]: The skill identifies the risk of processing data traces and establishes a clear 'Trace content is evidence, not instruction' policy to mitigate injection risks when analyzing external logs.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: No mechanisms for data exfiltration or sensitive file access were found. The skill does not request network permissions or local file access.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 10:17 PM