Your First Flow

This page runs two flows from the repository's samples/ directory: one driven by a schedule, and one you call directly from the command line. Together they show the two ways a flow receives work. You need the bin/octo binary from Installation.

A scheduled flow

samples/hello-world.yaml is the smallest complete config: one connector, one named processor, and one flow.

service:
  name: hello-world

connectors:
  - name: ticker
    type: cron

processors:
  - name: greeter
    type: log
    settings:
      level: info
      message: '"hello world! the date is " + body.date'

flows:
  - name: greet
    source:
      connector: ticker
      type: cron
      settings:
        schedule: "0,30 * * * * *"
        payload: '{"date": string(now)}'
    process:
      - ref: greeter

Reading it top to bottom:

• connectors declares a cron connector named ticker. Connectors own the machinery a flow needs to talk to the outside world.
• processors defines a reusable, named block. greeter is a log block whose message setting is a CEL expression — it can read the message body, vars, eventID, and correlationID.
• flows binds it together. The source attaches the flow to the ticker connector: a six-field cron schedule (with seconds) fires at seconds 0 and 30 of every minute, and the payload CEL expression — with now bound to the fire time — becomes each message's body. The process chain has one step, which references the named processor with ref: greeter.

Run it:

bin/octo run --config samples/hello-world.yaml

run is the default subcommand, so bin/octo --config samples/hello-world.yaml works too, and flags accept one or two dashes. The runtime logs its startup, prints a ready banner, then logs a greeting every 30 seconds:

time=... level=INFO msg="starting runtime" version=0.4.2 connectors=1 flows=1🚀  octo v0.4.2 is up and ready to roll!time=... level=INFO msg="hello world! the date is 2026-07-09T12:00:00Z"time=... level=INFO msg="hello world! the date is 2026-07-09T12:00:30Z"

Stop it with Ctrl-C.

Invoking a flow directly

Not every flow needs a schedule or a port. A flow with no source: gets an implicit source and becomes callable by name. samples/hello-invoke.yaml is exactly that:

service:
  name: hello-invoke

flows:
  - name: greet
    process:
      - type: set-payload
        name: build-greeting
        settings:
          value: '{"greeting": "hello, " + body.name + "!"}'

The set-payload block replaces the message body with the result of its value expression. Call the flow with octo invoke, passing the request body as JSON:

bin/octo invoke --config samples/hello-invoke.yaml --flow greet --data '{"name":"Ada"}'

{"event_id":"5c12a0e3f47b4d19a6c8f0b21d3e4a97","body":{"greeting":"hello, Ada!"}}

What prints is the whole result message, not just the payload: its event_id, the variables the flow set (none here), and the body. Pipe it through jq '.body' when the body is all you are after.

When --data is omitted, invoke reads the body from stdin, so you can pipe into it:

echo '{"name":"Grace"}' | bin/octo invoke --config samples/hello-invoke.yaml --flow greet

In invoke mode no sources are started — nothing binds a port or fires a schedule; only the requested flow runs, and its result message prints as JSON. This makes invoke the cleanest way to test a flow, and sourceless flows are also the building block for flow composition with flow-ref.

Next steps

Continue to the HTTP Quickstart to build a flow that serves real requests, or read Core Concepts for the full model behind flows, sources, and blocks.