skills/udecode/plate/major-task/Gen Agent Trust Hub

major-task

Pass

Audited by Gen Agent Trust Hub on Apr 4, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: Instructions specify running project-level install or setup commands when necessary for task execution.
  • [EXTERNAL_DOWNLOADS]: The agent is directed to git clone external library or framework repositories if they are not found locally.
  • [DATA_EXFILTRATION]: Uses gh (GitHub CLI) to post comments and update Pull Requests, which involves sending data to GitHub's servers as part of normal task synchronization.
  • [PROMPT_INJECTION]: The skill ingests untrusted data from issue trackers (GitHub, Linear), creating a surface for indirect prompt injection.
  • Ingestion points: Data is fetched from tracker descriptions, comments, and attachments using gh issue view, gh pr view, or Linear integrations.
  • Boundary markers: Use of <task> tags for argument passing provides minimal separation; no specific sanitization is defined for tracker content.
  • Capability inventory: The skill allows file access, shell command execution (e.g., git, gh, install), and network interaction.
  • Sanitization: There are no instructions for sanitizing or validating the content retrieved from external task trackers.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 4, 2026, 01:31 AM