agents-in-the-team
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The provided Python scripts,
agent_delegation_scorer.pyandagent_delivery_metrics.py, are deterministic and rely exclusively on the Python standard library. They do not perform any network operations, privilege escalation, or dynamic code execution. - [SAFE]: The skill incorporates a strong security posture, explicitly identifying threats like prompt injection and secret exposure. It provides actionable controls, such as limiting agent identity permissions, requiring human review for critical files (CI config, dependencies), and treating all external data as untrusted input.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of JSON backlog items and delivery records. While this constitutes an attack surface, the scripts use safe JSON parsing (
json.loads) and perform only arithmetic and string formatting. They do not execute or evaluate any content from the ingested data, effectively mitigating the risk of instruction injection.
Audit Metadata