Stuck on a problem and need help troubleshooting a discrepancy where the POST /api/v2/routing/skillgroups/{id}/agents endpoint reports 50 agents, but the Architect flow shows only 30 available for Predictive Routing. The integration is using the latest Platform SDK.
{
“code”: “bad_request”,
“message”: “Skill group capacity exceeds available agent count.”
}
The agents are online and assigned correctly in the admin UI. Is there a hidden filter in the predictive queue logic?
If I remember correctly, this mismatch stems from SIP registration states overriding UI status.
Cause: Agents marked online but with failed SIP credentials are excluded from predictive routing capacity.
Solution: Verify trunk health in Admin. Check for 401/407 errors on the specific BYOC trunks. Reset credentials if stale. Ensure availability_code matches the routing filter exactly.
What’s happening here is that load tests often miss the availability_code filter, causing agents to appear online but excluded from predictive routing capacity. Check the agent status in the API response to ensure it matches the routing filter exactly.