See what keeps going wrong for an agent
The goal's misses grouped into problems by rule: the kind of miss (a closed list: misunderstood, missing information, could not act, stuck, out of scope, handed off, the contact left, not reached, other) and the step of the agent where it went wrong. Each problem has a name built from those and your own step label, how many misses and different contacts it touched in the window, and a trend. A contact's words never reach a name or a count, so a builder may read it.
Requires an API key with the workspace:read scope.
apiKeyAuthorizationBearer <token>A secret API key from the dashboard (Developers → API keys): vl_live_…, or vl_test_… for test mode. Each operation says which scope the key needs.
id*stringThe agent id.
since?stringCount misses after this ISO timestamp. Default: the policy window (30 days).
date-timestatus?stringOnly active problems (a miss in the window) or quiet ones.
"active""quiet"The problems: a name, how many misses and contacts, and the trend; no conversation text
application/json- response
problems*array<>curl "https://api.vlayers.ai/v1/agents/$AGENT_ID/problems" \ -H "Authorization: Bearer $VOICELAYER_API_KEY"{ "problems": [ { "id": "0192f5fd-1625-7073-9c23-456789abcdef", "agentId": "0192f5c2-6f7a-7b9c-8d1e-2f3a4b5c6d78", "miss": "missing_information", "missStep": "check_availability", "name": "Did not have the answer at 'Check availability'", "status": "active", "firstSeenAt": "2026-10-20T15:04:05.000Z", "lastSeenAt": "2026-10-20T15:04:05.000Z", "members": 14, "distinctContacts": 9, "trend": { "recent": 8, "previous": 6 }, "byChannelClass": { "immediate": 9, "asynchronous": 5 }, "openProposalId": "0192f5fe-2736-7184-8d34-56789abcdef0" } ]}