Runtime guardrails for TypeScript AI agents. Prevents duplicate tool calls, enforces per-user cost budgets, and gates irreversible actions. MIT.
mcp circuit-breaker deduplication cost-control human-in-the-loop ai-agents idempotency guardrails approval-workflow llm langchain vercel-ai-sdk agentic-ai tool-calling mastra openai-agents agent-safety agent-authorization risk-gate
-
Updated
Aug 7, 2026 - TypeScript