troubleshooting

name	troubleshooting
description	Use when a Cognigy agent returns empty responses, a tool call or create_ai_agent fails, a resource is not found, setup_llm fails, or you need to diagnose a Cognigy MCP problem.

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
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

create_ai_agent failed

• The tool auto-rolls back created resources on failure. Safe to retry.
• "Could not find entry node": transient issue, retry immediately
• Endpoint step error: check project exists and is accessible

"Resource not found" errors

• All IDs are 24-char hex strings (e.g., 507f1f77bcf86cd799439011)
• UUIDs (36-char with dashes) are referenceIds — most tools need _id, not referenceId
• Use list_resources to find valid IDs

setup_llm fails

• See the llm-providers skill for valid provider and model strings
• Verify API key has access to the specified model

delete_resource fails

• Verify the resource ID is a 24-char hex string (not a referenceId UUID)
• Use list_resources to confirm the resource exists before deleting
• Deleting a flow may cascade-delete child nodes