agent-tool-definition

You were invoked as a slash command. The user's input:

$ARGUMENTS

Use that input to fill this prompt's variables (take the main content, topic, or task from it; ask only if a required value is missing and not supplied), then follow the prompt exactly.

Write a tool definition for an AI agent.

Tool: {tool} Inputs and output: {params}

Produce:

1. Name: a clear verb-noun name (search_orders, send_email) that says what it does.
2. Description: the most important part, because the agent chooses tools by reading it. State what it does, WHEN to use it, and when NOT to (to disambiguate from similar tools). Write it as instruction to the model, with an example use case.
3. Parameters: each with a clear name, type, whether required, a description with format and an example ("ISO date, e.g. 2026-01-15"), and an enum for closed choices. Few parameters, named plainly: every parameter description is a prompt the model reads.
4. Return shape: what the tool gives back, so the agent knows what to expect.
5. Error behavior: what errors it returns and how they should read to the agent ("date_range exceeds 90 days: split into smaller ranges", not a bare code): the error message is the agent's recovery instruction.

Rules: the description and parameter docs ARE prompts; write them as carefully as the system prompt. Size the tool at task granularity (one coherent capability), not one-per-endpoint. Make misuse hard through enums, required fields, and clear descriptions. Provide it in the format the target platform uses (JSON Schema for tool/function calling) if I name the platform, otherwise as a clear structured spec.