skills/lvlup-sw/exarchos/synthesize/Gen Agent Trust Hub

synthesize

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes shell-based version control operations (git), project testing scripts (npm), and GitHub CLI interactions (gh). These are legitimate actions for a PR management tool.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes project content and metadata to generate pull request descriptions.\n
  • Ingestion points: PR content is derived from branch data and feature IDs in SKILL.md and references/synthesis-steps.md.\n
  • Boundary markers: Uses quoted heredoc syntax (<<'EOF') when writing to temporary files to prevent shell variable expansion.\n
  • Capability inventory: Includes file system access, network-based VCS operations via MCP, and local command execution for tests.\n
  • Sanitization: Employs a specific validate_pr_body action to check PR descriptions against defined templates before submission.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 12:27 AM
Security Audit — agent-trust-hub — synthesize