Skip to content
Ahura OS
Get $10 free
Try Ahura OS with Get $10 trial credit. Register and start building.
All services
Coordinate

Ahura Streams

Real-time messaging between agents.

Publish and subscribe to events. Agents, workflows and your own systems react to each other as things happen, with routing handled by the platform.

What it does

  • Pub/sub topics

    Publish an event once and every subscriber gets it.

  • Event-driven workflows

    Start a workflow from an event, a schedule or another agent.

  • Filtering

    Subscribe to only the events that matter.

  • Ties the platform together

    Jobs, partitions and billing all emit events you can listen to.

An agent calls it like this

{ "tool": "streams.publish", "topic": "reports.ready", "data": { "id": "r_91" } }

Interactive simulation

Try Ahura Streams

Press Run to send the call. The dashboard shows what it would use in a partition. Simulated: nothing is actually executed.

ahura mcp · ptn_demo_7f3a

# Announce that a report is ready.

→ POST /mcp

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "streams.publish",
    "arguments": {
      "topic": "reports.ready",
      "data": {
        "id": "r_91"
      }
    }
  }
}

# Another agent catches up.

→ POST /mcp

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "streams.read",
    "arguments": {
      "topic": "reports.ready",
      "after": "ev_00930",
      "limit": 10
    }
  }
}

Ready. Press Run.

ptn_demo_7f3a

Partition · N. Virginia · created just now

Simulated
Activeus-eastRetention hold
Run time
0.0s
0 calls
Size
2.3MB
0.00 ACU used

Services in this partition

  • Streams

    0 calls · 0.00 ACU

Spend by service

Ledger

  • No usage yet.

Why it matters

Where Ahura Streams beats running a message broker.

Agents work best when they can react to each other. Brokers make that possible but need running, securing and watching.

A broker you operate

  • Run and scale a broker, and manage its access.
  • Build durable logs, retries and dead-letter handling yourself.
  • Poll for changes or write your own webhook sender and signer.
  • Add a scheduler for periodic work.

Ahura Streams

  • Publish and subscribe with a tool call.
  • A durable log, 24-hour retries and a dead-letter queue are included.
  • Signed webhooks and long-polling are built in.
  • Cron and rate schedules publish events by themselves.

  • Durable

    Catch up from any point after being offline.

  • Trusted

    Signed webhook deliveries and inspectable failures.

  • Connected

    Workflows start from events.

  • Aware

    Platform events such as payments and freezes arrive on the same bus.

Streams is for coordination, not for very high-throughput ordered data pipelines.

In depth

What Ahura Streams can do.

  • Topics inside a partition

    Publish structured events to named topics. Subscribers see only what you route to them.

  • A durable log

    Every event is kept, ordered by time, so a consumer that was offline can catch up from where it left off.

  • Read, or wait

    Read history with a cursor, or long-poll for the next event and wake the moment it arrives.

  • Pattern subscriptions

    Match events by their content with the same pattern syntax used for cloud event routing, validated when you subscribe.

  • Webhooks, signed

    Deliver to your URL with a timestamped signature you can verify. Failed deliveries retry for up to 24 hours, then land in a dead-letter queue you can inspect.

  • Schemas on topics

    Attach a JSON schema to a topic and bad events are rejected at publish time.

  • Idempotent publish

    Send an idempotency key and duplicates within 24 hours are ignored.

  • Schedules

    Create cron or rate schedules that publish an event on their own: an agent’s heartbeat.

  • Platform events

    Account and partition state changes, payments, signatures and workflow results all appear as events you can listen to.

  • Retention you set

    Choose how long each topic keeps events, up to the partition’s lifetime.

Every call, listed

12 tools your agent can call. Each one is a standard MCP tool with a typed schema and a price shown in tools/list.

Publish and read

streams.publish
Publish one event, or up to 10.
streams.read
Read events from a topic or the combined log, forwards or backwards.
streams.tail
Long-poll for new events.
streams.topics
Topics with counts and last activity.
streams.topic.configure
Set retention and a schema.

Subscriptions

streams.subscribe
Route matching events to a webhook, another topic or a workflow.
streams.subscriptions.list
See subscriptions.
streams.unsubscribe
Remove one.
streams.subscription.deliveries
Recent deliveries and failures.

Schedules and platform

streams.schedule
A cron or rate schedule that publishes an event.
streams.schedules.list / delete
Manage schedules.
streams.platform.read
Account-level platform events.

Delivery between topics is best-effort in order. Streams is for coordination and reaction, not for very high-throughput ordered pipelines.

Use cases with Streams

All use cases

and 4 more

Works with

Next: Ahura Memory

Ahura Identity

A verifiable identity for every agent.

Ahura Memory

Fast working memory for agents.

Ahura Store

Persistent files and objects.

Ahura NoSQL

Document collections without the setup.

Partitions

Where this service's data lives and how it freezes.