tailwind-impl-migration-v3-v4

Warn

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [METADATA_POISONING]: The skill frontmatter identifies the author as 'OpenAEC-Foundation', which contradicts the provided author context of 'impertio-studio'. This deceptive metadata obscures the actual origin of the skill.\n- [REMOTE_CODE_EXECUTION]: The migration guide utilizes the official @tailwindcss/upgrade utility via npx, which involves fetching and executing code from the npm registry. While from a well-known source, this represents a remote execution vector.\n- [COMMAND_EXECUTION]: The skill provides numerous shell commands for auditing project files (grep, find) and managing dependencies (npm install). These are standard operations for development and migration tasks.\n- [INDIRECT_PROMPT_INJECTION]: The agent is instructed to process user-controlled source code and templates during the migration. \n
  • Ingestion points: Local source files including .tsx, .vue, .svelte, .html, and .css files. \n
  • Boundary markers: Absent; the skill does not specify delimiters to separate untrusted code from instructions. \n
  • Capability inventory: Shell execution (npx, npm, git), file modification, and dependency management. \n
  • Sanitization: Absent; no sanitization or validation of the ingested code is recommended before processing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 23, 2026, 10:58 AM
Security Audit — agent-trust-hub — tailwind-impl-migration-v3-v4