Tool Definition & Schema Designer

Design a tool for an LLM agent. The capability: {{capability}}. Underlying API or function: {{underlying_api}}.

Provide: a clear tool name (verb_noun); a description written for the model that says when to use it, when NOT to use it, and what it returns; a strict JSON Schema for the parameters (enums over free text, sensible defaults, required vs optional, with a description on every field); safe-by-default behaviour (read-only mode, dry-run flag, limits); the success response shape and an error shape with machine-readable codes and a 'what to do next' hint the model can act on; pagination or truncation rules to protect the context window; two example calls; and the permission tier (read, reversible write, irreversible) with approval requirements. Point out ways this tool could be misused through prompt injection.