vercel-composition-patterns

Warn

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
  • [METADATA_POISONING]: The skill metadata in SKILL.md identifies the author as 'vercel', and the AGENTS.md file claims the content is 'vendored from Vercel'. However, the provided author context for this skill is 'akillness'. This discrepancy constitutes misleading metadata that impersonates a trusted organization.
  • [INDIRECT_PROMPT_INJECTION]: The skill's instructions require the agent to ingest and refactor external codebase data, creating a surface for potential injection.
  • Ingestion points: Component architecture and source code from external projects as described in SKILL.md.
  • Boundary markers: None. There are no instructions for the agent to distinguish between its guiding rules and potentially malicious instructions embedded in the target code.
  • Capability inventory: The agent typically has permissions to modify local files and execute development tools (npm/yarn/git) to apply these patterns.
  • Sanitization: Absent. The skill provides no validation or filtering mechanisms for the code it processes.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 14, 2026, 08:42 AM