Skip to main content
POST
/
threads
/
{thread_id}
/
runs
/
stream
Create And Stream Run
curl --request POST \
  --url https://api.example.com/threads/{thread_id}/runs/stream \
  --header 'Content-Type: application/json' \
  --data '
{
  "assistant_id": "<string>",
  "input": {},
  "config": {},
  "context": {},
  "checkpoint": {},
  "stream": false,
  "stream_mode": "<string>",
  "on_disconnect": "<string>",
  "multitask_strategy": "<string>",
  "command": {},
  "interrupt_before": "<string>",
  "interrupt_after": "<string>",
  "stream_subgraphs": false,
  "metadata": {}
}
'
{
  "error": "<string>",
  "message": "<string>",
  "details": {}
}

Documentation Index

Fetch the complete documentation index at: https://docs.aegra.dev/llms.txt

Use this file to discover all available pages before exploring further.

Path Parameters

​
thread_id
string
required

Body

application/json

Request model for creating runs

​
assistant_id
string
required

Assistant to execute

​
input
Input · object

Input data for the run. Optional when resuming from a checkpoint.

​
config
Config · object

Execution config

​
context
Context · object

Execution context

​
checkpoint
Checkpoint · object

Checkpoint configuration (e.g., {'checkpoint_id': '...', 'checkpoint_ns': ''})

​
stream
boolean
default:false

Enable streaming response

​
stream_mode

Requested stream mode(s)

​
on_disconnect
string | null

Behavior on client disconnect: 'cancel' (default) or 'continue'.

​
on_completion
enum<string> | null

Behavior after stateless run completes: 'delete' (default) removes the ephemeral thread, 'keep' preserves it.

Available options:
delete,
keep
​
multitask_strategy
string | null

Strategy for handling concurrent runs on same thread: 'reject', 'interrupt', 'rollback', or 'enqueue'.

​
command
Command · object

Command for resuming interrupted runs with state updates or navigation

​
interrupt_before

Nodes to interrupt immediately before they get executed. Use '*' for all nodes.

​
interrupt_after

Nodes to interrupt immediately after they get executed. Use '*' for all nodes.

​
stream_subgraphs
boolean | null
default:false

Whether to include subgraph events in streaming. When True, includes events from all subgraphs. When False (default when None), excludes subgraph events. Defaults to False for backwards compatibility.

​
metadata
Metadata · object

Request metadata propagated to OTEL trace attributes (langfuse.trace.metadata.<key>). Keys must match [A-Za-z0-9_-]{1,64}. Values must be primitive (str, int, float, bool); string values are capped at 512 characters. Maximum 32 keys. Use this for filterable attributes (tenant, feature flag, environment, sub-agent type) rather than payload data.

Response

Server-Sent Events stream