Skip to main content
POST
/
v1
/
triage
/
issues
/
{id}
/
agent-sessions
/
{session_id}
/
prompts
Prompt Norm
curl --request POST \
  --url https://api.bland.ai/v1/triage/issues/{id}/agent-sessions/{session_id}/prompts \
  --header 'Content-Type: application/json' \
  --header 'authorization: <authorization>' \
  --data '
{
  "detail": "<string>"
}
'
{
  "data": {
    "session_id": "0431f6ad-fdd3-4408-a6cc-909842d1db41",
    "run_id": "56cae586-0670-4f1b-8113-96a2421d56d5",
    "intent": "fix"
  },
  "errors": null
}

Documentation Index

Fetch the complete documentation index at: https://docs.bland.ai/llms.txt

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

​
Overview

Dispatches a Norm run against the issue. Asynchronous, returns 202 once queued. Progress, tool calls, and final output stream into the issue’s Activity feed as entry_kind: "agent" entries. The first word of detail is parsed as an intent. Slashes are optional.
  • /fix (default), full read+write toolbelt including pathway editing, tool config, and snippet/pipeline tests.
  • /verify, read-only. Norm checks the current draft against the evidence and reports.
  • /status, read-only. Norm summarizes current state, leading diagnosis, and any draft verification.
  • /debug, alias for /fix.
The @norm mention in Add Comment calls this same flow.

​
Headers

​
authorization
string
required
Your API key for authentication.

​
Path Parameters

​
id
string
required
Internal UUID of the issue.
​
session_id
string
required
Internal UUID of the agent session, from Create Agent Session or the latest_agent_session.id field on the issue.

​
Body Parameters

​
detail
string
required
The prompt to send to Norm. 1-4000 characters. Begin with /fix, /verify, /status, or /debug to set the intent. The intent prefix is stripped before the message reaches Norm.

​
Response

Returns 202 Accepted. The body acknowledges the dispatch but does not include Norm’s reply. Subscribe to the issue’s Activity feed (or refetch it on a poll) to read Norm’s progress and final output.
​
data.session_id
string
ID of the session the prompt was dispatched to. Matches the {session_id} in the URL.
​
data.run_id
string
ID of the Norm run created for this prompt. Each prompt produces one run; multi-turn conversations on the same session produce multiple runs.
​
data.intent
string
The parsed intent. One of fix, verify, or status. /debug is normalized to fix.
​
errors
null | array
null on success. Returns 404 if the issue or session does not exist, or if the session does not belong to the named issue.
{
  "data": {
    "session_id": "0431f6ad-fdd3-4408-a6cc-909842d1db41",
    "run_id": "56cae586-0670-4f1b-8113-96a2421d56d5",
    "intent": "fix"
  },
  "errors": null
}
Docs for agents: llms.txt