# ReqRun

ReqRun is an LLM request reliability layer for OpenAI-compatible non-streaming chat completion requests that need durable queueing, retries, idempotency, wait mode, and request status.

Important pages:
- Homepage: https://reqrun.com/
- Quickstart: https://reqrun.com/docs/quickstart
- API reference: https://reqrun.com/docs/api
- TypeScript SDK: https://reqrun.com/docs/sdk/typescript
- Request execution concept: https://reqrun.com/docs/concepts/request-execution
- Retry logic concept: https://reqrun.com/docs/concepts/retry-logic
- Idempotency concept: https://reqrun.com/docs/concepts/idempotency
- Why direct LLM API calls break in production: https://reqrun.com/blog/direct-llm-api-calls-break-production
- Idempotency for LLM requests: https://reqrun.com/blog/idempotency-for-llm-requests
- Wait mode and async fallback: https://reqrun.com/blog/wait-mode-and-async-fallback
- Security: https://reqrun.com/security
- Pricing: https://reqrun.com/pricing
- Terms: https://reqrun.com/terms
- Privacy: https://reqrun.com/privacy
- Refund Policy: https://reqrun.com/refund-policy

Current product boundary:
- OpenAI-compatible POST /v1/chat/completions
- GET /v1/requests/{id}
- OpenAI-only
- Non-streaming only
- Durable SQLite-backed queue for v1
- No multi-provider routing
- No streaming
- No general workflow builder
- No raw payload logging in normal logs

Primary SDK import:
import { ReqRun } from "@reqrun/sdk";