The Agent Skills Directory

[SAFE]: The skill serves as an educational resource for Vercel AI SDK v6 development. All code examples use standard SDK patterns and reference official documentation.
[PROMPT_INJECTION]: The skill describes agent patterns that ingest user-provided messages and use tools with parameters derived from model outputs, which constitutes a surface for indirect prompt injection.
Ingestion points: The runAgent function in rules/agent-multi-step.md accepts a message string directly from the user context.
Boundary markers: The examples do not include explicit delimiters or system instructions to ignore embedded commands within the user input.
Capability inventory: Example tools such as lookupUser demonstrate capabilities like database access (db.users.findUnique).
Sanitization: The provided code snippets do not include input validation or sanitization logic, relying on the underlying LLM's safety filters.

ai-sdk