# Wrong tool use

The agent skips, ignores, fabricates or over-uses a tool, or picks the wrong one.

**Is a:** [Failure kind](/reference/ontology/unattended-agent-failures/failure-kind)

**Guarded by:** [Descriptive tool names](/reference/ontology/unattended-agent-failures/descriptive-tool-names)

**Can cause:** [Gaming the check](/reference/ontology/unattended-agent-failures/gaming-the-check), [Wrong goal](/reference/ontology/unattended-agent-failures/wrong-goal)

ToolFailBench labels tool-use failures Tool-Skip, Result-Ignore, Output-Fabrication and Unnecessary-Tool-Use; its authors report the best of 19 models at 86.33% clean tool use ([ToolFailBench](https://arxiv.org/abs/2607.04686), from its abstract). Tool choice also follows how a tool is described: BiasBusters finds semantic match between query and tool description is the strongest interpretable driver across seven models ([BiasBusters](https://arxiv.org/abs/2510.00307)), and ToolTweak raised a tool's selection rate from about 20% to as high as 81% by editing only its name and description ([ToolTweak](https://arxiv.org/abs/2510.02554)). Models also follow social cues such as authority or consensus more than formal role ([Who is In Charge?](https://arxiv.org/abs/2510.01228)).

Our own reading: a fabricated tool result is a false input the later checks may not see, so this can cause [gaming the check](/reference/ontology/unattended-agent-failures/gaming-the-check) or a [wrong goal](/reference/ontology/unattended-agent-failures/wrong-goal) downstream. Guard: [descriptive tool names](/reference/ontology/unattended-agent-failures/descriptive-tool-names).

---

Published 2026-10-29. Web page: https://www.mackenziebowes.com/reference/ontology/unattended-agent-failures/wrong-tool-use