AI CostGuard, a local-first runtime safety layer for AI agents, prevents runaway costs and budget explosions before API calls execute.

AI CostGuard wraps OpenAI-compatible clients and function-style SDK calls, estimates request cost locally, blocks budget overruns, detects repeated prompts, and emits structured events without requiring a SaaS control plane, cloud dashboard, proxy gateway, telemetry service, or billing reconciliation service.

The tool blocks requests before the provider method is called when they trigger runtime conditions: UNKNOWN_MODEL, BUDGET_EXCEEDED, MAX_STEPS_EXCEEDED, LOOP_DETECTED, or RETRY_STORM_DETECTED.

AI CostGuard is installed via npm and includes a local-only dashboard that reads JSONL event logs, CLI checks, integrations with OpenAI SDK, Anthropic SDK, Vercel AI SDK, LangChain, Mastra, and CrewAI, and an optional Redis-backed Pro helper for shared spend tracking.