connections-optimizer

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses browser control and desktop automation (Apple Mail) to facilitate network analysis and drafting. These operations are strictly limited to the stated purpose of network reorganization and outreach preparation.
  • [DATA_EXFILTRATION]: While the skill accesses sensitive social graph data and connection inventories, it lacks any instructions to send this data to unauthorized external endpoints. All drafted content and prune queues are presented to the user for review before any action is taken.
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting external profile data from social media and research results. This is mitigated by the explicit 'review-first' policy and the structured 'Review Pack' format which ensures a human-in-the-loop before any automated execution or outbound communication.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 06:50 AM
Security Audit — agent-trust-hub — connections-optimizer