troubleshooting

Troubleshooting

Agent returns empty response

1. Inspect the agent flow and tools first:
   • list_resources { resourceType: "tool", aiAgentId }
   • duplicate toolId values can cause failed tool execution and empty responses — this is a flow issue, not an LLM or connection issue; remove the duplicate rather than reconfiguring the LLM
2. Check LLM exists: list_resources { resourceType: "llm_model", projectId } If none: run setup_llm
3. Check agent description is not empty: get_resource { resourceType: "agent", id }
4. Check endpoint is connected: get_resource { resourceType: "endpoint", id } Verify flowId is set and URLToken exists
5. Endpoint JUST created (seconds ago)? Endpoint config propagates briefly, and a session whose FIRST message hit the stale config stays cached as broken — wait a few seconds and retry with a NEW sessionId, not the same one.
6. LLM Prompt flows: the assigned LLM must have a connection (connectionId set on the llm_model). A connectionless LLM fails silently — the node's default error handling is "continue" with an empty message, which looks like an empty response.

talk_to_agent fails (any HTTP error or timeout)

talk_to_agent sends through Cognigy Endpoint Test Mode (/test/<token>) so test messages are not counted as billable. It never re-sends a failed message on the regular (billable) endpoint by itself; a failure comes back as error with testMode, endpointUrl and status-aware _hints.

Step 1, always: find out whether the message was processed. An HTTP error does not prove it was not. A REST endpoint's Execution Finished transformer runs after the flow and can return any status (a 404 included), and a gateway timeout can hide a completed execution. Waiting or reusing the sessionId does not prevent a duplicate. Check:

• get_resource { resourceType: "conversation", id: "<sessionId>" } for the transcript (needs the endpoint to collect conversations), or
• continue the same sessionId with a neutral follow-up ("what did you just do?"), or verify the side effects of the agent's tools.