authoring-signals-scouts

Pass

Audited by Gen Agent Trust Hub on Jun 15, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The documentation includes extensive instructions on identifying and mitigating indirect prompt injection. It guides authors to treat all external data, such as warehouse rows or repository code, as untrusted input that should never be interpreted as instructions.
  • [COMMAND_EXECUTION]: The skill mentions the use of CLI tools like git, node, npx, and Django management commands (manage.py) within the context of scout development and testing. These are presented as standard tools for the described workflow.
  • [EXTERNAL_DOWNLOADS]: The guide references patterns for fetching external data using git clone and curl. It acknowledges the potential security risks and provides mitigation strategies, such as using shallow clones and ensuring content is treated as data for analysis only.
  • [DATA_EXFILTRATION]: The skill outlines the process for emitting findings to an internal inbox and provides specific instructions on sanitizing personally identifiable information (PII) to prevent accidental data exposure.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 15, 2026, 12:36 PM
Security Audit — agent-trust-hub — authoring-signals-scouts