Glossary
Tool Sandboxing
Tool sandboxing restricts what an agent's tools can access or modify (for example via allowlists, timeouts, and isolated environments) to reduce the blast radius of mistakes or attacks.
Plain-English meaning
In this game, Tool Sandboxing is used as a vocabulary card for recognizing how market and technology concepts fit together. The short idea is: constrain tool permissions.
The term is not shown as a recommendation. It is included so players can learn the language they may see in exchange interfaces, wallet prompts, research notes, AI product pages, or on-chain analytics dashboards.
Why it belongs with AI Agents and Evaluation
Evaluation and controlled tool use help teams improve reliability without assuming the model is always correct or safe.
When solving the puzzle, compare the job this term performs with nearby cards. A correct group usually shares a function, risk type, workflow, or market structure rather than simply sharing similar wording.
Where you might see it
You might encounter this term while reading educational explainers, product documentation, risk disclosures, market dashboards, or beginner guides. Always separate vocabulary learning from financial decision-making.