A durable message queue,
in one API call.
Send and receive messages with at-least-once delivery, visibility timeout, retries and built-in dead-letter — on your flat plan. The agent-native, no-lock-in alternative to AWS SQS, built on portable Postgres.
A queue shouldn’t need IAM policies and a per-request bill
Background jobs, fan-out and decoupling services all need a durable queue — but SQS means IAM policies, the console, a separate DLQ with a redrive policy, and a per-request bill that’s hard to predict. koigrid Queues give you the same model — send, receive with a visibility timeout, ack to delete, at-least-once redelivery, and dead-letter after maxReceives — from one API call and a scoped token. Concurrent consumers never get the same message twice (FOR UPDATE SKIP LOCKED). It runs on portable Postgres, so there’s no AWS-specific lock-in.
How koigrid compares
| koigrid | AWS SQS | |
|---|---|---|
| Pricing | Flat (included) | Per request + transfer |
| Delivery | At-least-once + visibility timeout | Same |
| Dead-letter | Built in (maxReceives) | Separate DLQ + redrive policy |
| No double delivery | Yes (skip-locked) | Yes |
| Setup | One API call / your agent | IAM, policies, console |
| Lock-in | None (Postgres model) | AWS-specific API + IAM |
From public docs, 2026. FIFO ordering and long-polling are on the roadmap.
What you’d pay elsewhere
50M messages/month with dead-letter and alerting
koigrid Queues (Starter)
€12/mo/mo
flat — queues + messages included in your plan
AWS SQS
~$20+/mo/mo
per-request pricing + data transfer that grows with volume
flat & predictable
Prices are indicative; koigrid is flat-rate and EU-hosted.
Everything you need to decouple services
Send with delay & dedupe
Delay a message up to 15 minutes; pass a dedupe id to make sends idempotent.
Visibility timeout
A received message is hidden while you process it; ack to delete, or it reappears — at-least-once by design.
Dead-letter built in
Set maxReceives on the queue — a poison message stops being redelivered after N receives. No separate DLQ to wire.
No double delivery
Concurrent consumers claim different messages (FOR UPDATE SKIP LOCKED) — safe to scale horizontally.
Agent-native
Create a queue and send/receive/ack from Claude Code or any LLM via the API + a scoped token.
No lock-in
Built on portable Postgres — the model moves with you, unlike the AWS-specific SQS API.