database-migrations

Pass

Audited by Gen Agent Trust Hub on Nov 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to identify and execute shell commands for database migrations, including repository-specific wrappers (e.g., make, task) and standalone tools (e.g., goose, flyway, alembic). It also uses wrangler d1 commands for Cloudflare-specific migrations.
  • [EXTERNAL_DOWNLOADS]: The instructions frequently prescribe the use of npx wrangler, which fetches the wrangler package from the npm registry.
  • [REMOTE_CODE_EXECUTION]: Use of npx results in the execution of code downloaded from a remote package registry at runtime.
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted repository files to determine the migration workflow, creating a potential surface for indirect prompt injection.
  • Ingestion points: SKILL.md and references/cloudflare-d1-migrations.md instruct the agent to read wrangler.toml, package.json, Makefile, and existing migration SQL files to detect the repository's "migration contract."
  • Boundary markers: No specific boundary markers or instructions to disregard embedded malicious content are defined when reading these configuration and script files.
  • Capability inventory: The agent has the capability to execute shell commands (via wrangler, make, etc.) and write new files to the local directory.
  • Sanitization: The skill lacks explicit instructions to sanitize or validate strings parsed from repository configuration files before they are interpolated into generated shell commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 15, 2026, 05:15 PM