notifications-manager

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the gh api command-line tool to perform actions such as listing, filtering, and marking notifications as read. This is the intended purpose of the skill and does not involve arbitrary command execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests untrusted data (notification titles, repository names, and issue/PR details) from the GitHub API.
  • Ingestion points: Notification data fetched via gh api and presented to the user/agent.
  • Boundary markers: None explicitly defined in the skill instructions to separate notification content from agent instructions.
  • Capability inventory: The skill can perform batch operations like marking all read or unsubscribing from threads based on the processed content.
  • Sanitization: No explicit sanitization or validation of the notification text is mentioned in the skill definition.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 05:34 AM
Security Audit — agent-trust-hub — notifications-manager