Test the AI agent
Chat with an agent as a test customer (works for inactive agents too) and get its reply, tools and knowledge used. Uses AI credits.
bearer testingAuthorizationBearer <token>API key (rsl_live_…) from Settings → Developers.
Scope: testing
agentId*stringAI agent id. Never invent ids – list resources first.
1 <= length <= 64application/json- body
message?stringWhat the test customer writes. Omit to only wait for a pending reply.
1 <= length <= 5000testSessionId?stringContinue a previous test conversation (from the last result).
length <= 2000waitSeconds?integerHow long to wait for the AI reply.
0 <= value <= 5545Success
application/json- response
status*stringreplied = AI answered; handed_off = AI escalated to a person; no_reply = AI chose not to answer; pending = still thinking, call again with testSessionId and no message.
"replied""no_reply""handed_off""error""pending"replies*array<string>toolsUsed*array<string>knowledgeUsed*array<string>testSessionId*stringconversationId*stringdashboardUrl*stringcurl -X POST "https://example.com/agents/string/test" \ -H "Content-Type: application/json" \ -d '{}'{ "status": "replied", "replies": [ "string" ], "toolsUsed": [ "string" ], "knowledgeUsed": [ "string" ], "testSessionId": "string", "conversationId": "string", "dashboardUrl": "string"}Evaluate a test conversation POST
Evaluate a test conversation (from agents.test) against plain-language criteria with an AI judge. Used by `resolido test`. Uses AI credits.
Update AI agent PATCH
Update an AI agent's name, instructions, model, domains, signature and behaviour. Only sent fields change. Owner/admin.