Agents

The agent responds to users, works with connected knowledge base materials, calls authorized tools, and can help the user navigate the interface.

This section covers the full agent lifecycle: create, train, test, publish, connect tools, and understand why an answer turned out the way it did.

Main tasks

• Creating an agent manually - basic agent settings.
• Draft, publication and launch of the agent - how to bring the agent to a working state.
• Agent testing - how to check responses before launch.
• Knowledge base sources and agent rights - which materials are connected to the agent.
• Agent Capabilities and Tools - which capabilities and tools are available to the agent.
• Project Operator and Agent Replies - how to prevent team messages from triggering the agent and what happens to a response already being prepared.

Recommended order

1. Create an agent manually or through the auto-training wizard.
2. Add the role, response rules, and knowledge base sources.
3. Check the agent permissions and allowed tools.
4. Test answers on real questions.
5. Publish the draft and connect the agent to the required channel.

What to check when something is wrong

If answers differ from what was expected, start with the published agent version: check the instruction, knowledge sources, variables, MCP calls, and tool permissions. If the question is related to the interface, open the article for the required section and compare the steps with the current screen.