notifications
Pass
Audited by Gen Agent Trust Hub on Mar 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the
gh-managercommand-line tool to list and mark GitHub notifications as read. This tool is a vendor-provided resource for managing repository interactions. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from GitHub notifications (such as issue titles, comments, and mentions) provided by external users.
- Ingestion points: Data enters the agent context via the
gh-manager notifications listcommand (SKILL.md). - Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the processed notification content.
- Capability inventory: The skill has the capability to modify the user's GitHub state by marking notifications as read via
gh-manager notifications mark-read(SKILL.md). - Sanitization: No validation or filtering is performed on the notification text before it is presented to or processed by the agent.
Audit Metadata