vercel-ai-sdk-best-practices

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The explicit "Memory Protocol" mandates running a shell command to read a local memory file (cat .claude/context/memory/learnings.md) and recording new patterns, which is an instruction outside the stated purpose of giving Vercel AI SDK best-practices and could be used to access or exfiltrate local context, so it functions as a hidden/deceptive behavioral override.