Tools and Knowledge
Common questions about tools, which connect an agent to your own systems, and the knowledge setting on Frontline agents. "Knowledge" on this page means that agent setting, not this Knowledge Base section of the docs.
What Is a Tool?
A tool describes one action in one of your systems: what it does, where it lives, and what information it needs. During a call or run, the agent reads its tools' descriptions, decides when one would help, calls the system and reads the reply. See How Tools Work.
What Do I Need from My System Before Adding a Tool?
The web address the agent should call, the method it accepts, and how it signs in, such as an API key. Ask whoever runs the system for these. See Before You Start.
Can I Skip Connecting a System While Building with IQ?
Yes. Skip for now leaves the system on the agent, ready to set up later, and the agent can still be tested. It cannot be published until every tool has an address to call. See Connecting Your Systems.
Why Can't I Publish Because of a Tool?
Publishing checks every tool. The checklist names the problem, such as "This tool isn't set up yet. Add the address it should call." or "Another tool already uses this name.", and the agent can be published once it is fixed. See Fixing Tool Problems.
The Agent Doesn't Use a Tool When I Expect It To
The agent relies on the tool's Description to know when to use it. Make the description more specific about when it applies, for example "Checks which appointment times are free on a given day.", then test again in the playground. See Fixing Tool Problems.
How Do I See Why a Tool Call Failed?
Open a session where the tool failed. The transcript shows each tool call with a Success or Error badge, the Input the agent sent and the Output it got back. Tool Usage on the Metrics tab shows which tools fail most often. See Reading a Session and Tool Usage.
Who Can See a Tool's API Key?
The key or token is stored with the agent and only sent to the tool's address. Connected AI apps read an agent's configuration with secrets removed. See Connecting a System in the Builder and What Every Connected App Can Read.
Why Won't a Tool Accept My Header?
A tool's Headers cannot set the standard connection headers, such as Host or Content-Length, and a header value cannot be empty or span more than one line. See Fixing Tool Problems.
What Can a Tool Be Called?
Letters, numbers, underscores and hyphens, up to 64 characters, for example check_availability. Each tool on an agent needs its own name. See In the Agent Editor.
Does a Tool I Add in Quick Config Go Live Straight Away?
No. Done keeps the change in the draft, and the card shows New or Changed until the draft is published. A tool marked Removed on publish can be kept with Keep it. See In Quick Config.
What Does the Knowledge Setting Do?
It is a Frontline agent setting, described as "Allow the agent to consult the configured knowledge base.", and it is on by default. It appears as Knowledge under Recording & knowledge in the builder and as Knowledge base agent in the editor's Model section, and both switches change the same setting. See Giving an Agent Knowledge.
Do Background Agents Have the Knowledge Setting?
No. To give any agent facts it needs, such as opening hours or policies, put them in its instructions, or give it a tool that looks them up. See Giving an Agent Knowledge.