Executes an agent using its currently deployed revision with streaming output.
Returns results as Server-Sent Events (SSE) for real-time progress updates.
API-Only Endpoint
This endpoint is available via API only and requires a Fruxon bearer token: Authorization: Bearer YOUR_TOKEN.
Fill in your parameters below and the code samples will update automatically.
To generate a token, see the Settings guide.
Python SDK Available
You can also call this endpoint using the Fruxon Python SDK.
Install with pip install fruxon — FruxonClient.execute(), .execute_revision(), and .stream() cover this surface, and the fruxon run CLI wraps them with --stream / --no-stream / --revision.
Authorization
Token Fruxon token presented as a Bearer token: Authorization: Bearer fx_pat_… (personal) or fx_sat_… (service account). Generate one from Settings → API Access.
In: header
Path Parameters
The unique identifier of the agent.
The tenant identifier
The execution parameters and inputs.
Key-value parameters to pass to the agent
File attachments for the execution. Each entry references a file already uploaded through
POST /storage; they are attached to the run's user turn, so a vision-capable model
receives the images directly and every file stays addressable by its
<fileType:fileId/> link. Maximum 20.
Optional session ID for conversation continuity
Integration-config variant this execution resolves against. Honoured on the draft-test
endpoints; :execute / :stream run a deployed revision and always resolve
Production configs, so the value is ignored there.
"UNSPECIFIED" | "PRODUCTION" | "SANDBOX"Optional environment slug for per-end-customer cost tracking and analytics. Auto-creates the environment on first use (subject to the per-tenant environment cap).
length <= 64How to route a connection the agent uses during this test, by connection id: SIMULATED,
READ_THROUGH, or NONE — shown as Live, which reaches the connection with its production
credentials. Honoured on the draft-test endpoints only, and it makes the test a sandbox run. A sandbox test
simulates every connection left out that has no sandbox saved; nothing is saved on the connection.
The reply style to test with instead of the agent's active one: {"version": 3} for a particular version
(a draft under review, say), or {"off": true} for none — send the same message both ways to compare.
Honoured on the draft-test endpoints only; a deployed run always gets the active reply style.
Response Body
curl -X POST "https://api.fruxon.com/v1/tenants/string/agents/string:stream" \ -H "Content-Type: application/json" \ -d '{}'"string"{
"type": "string",
"title": "string",
"status": 0,
"detail": "string",
"instance": "string",
"property1": null,
"property2": null
}{
"type": "string",
"title": "string",
"status": 0,
"detail": "string",
"instance": "string",
"property1": null,
"property2": null
}Executes an agent using its currently deployed revision.
Runs the agent synchronously and returns the complete result.
Gets the agent's conversations and escalations bucketed over time.
A conversation is a customer visit, counted the way the Application's support KPIs count it: the customer's messages in one conversation, split wherever they went quiet for four hours. It is escalated when the agent handed it to a person, or tried to and reached nobody (`deadEnded`). Production traffic only. Each conversation is counted in the bucket it started in; buckets with none are omitted. Counting is live over the workspace's run history, so a window reaching further back is counted from `countedFromMs`.