software-migration-engineer

Warn

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The documentation in references/integrations.md instructs users to download and execute multiple Node.js packages from the npm registry using npx. While some packages originate from well-known entities, others like @AshDevFr/pusher-channels-mcp-server are from unverified third-party authors.
  • [REMOTE_CODE_EXECUTION]: The skill facilitates remote code execution by providing commands to install and run external MCP servers via npx -y. This pattern downloads and executes code directly from a remote registry.
  • [COMMAND_EXECUTION]: The skill explicitly suggests using shell commands through Bash as a fallback for migration tasks, including tools like terraform, kubectl, and platform-specific CLIs (gh, glab, az), which could be vulnerable to manipulation if provided with malicious input from analyzed codebases.
  • [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to ingest and analyze untrusted external data, including repository structures, changelogs, and project management issues. There are no defined boundary markers or sanitization procedures mentioned to mitigate the risk of embedded instructions in these sources affecting the agent's behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 13, 2026, 01:22 PM
Security Audit — agent-trust-hub — software-migration-engineer