tool-design

Design agent-facing tool interfaces with consolidated schemas, structured descriptions, and recoverable error messages.

Updated Apr 7, 2026
One-click install
npx skills add https://github.com/NT-boop-star/BRMV-tract --skill tool-design-nt-boop-star
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tool-design
Source: https://github.com/NT-boop-star/BRMV-tract/tree/main/antigravity/skills/tool-design
Command: npx skills add https://github.com/NT-boop-star/BRMV-tract --skill tool-design-nt-boop-star

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Agents select and call tools based solely on descriptions, so vague names, ambiguous parameters, and unhelpful error messages cause tool-selection failures that prompt engineering cannot fix. This Skill provides principles, templates, and utilities for designing tool interfaces that agents can use correctly. ## Core Features & Use Cases - Tool Description Engineering: Structure descriptions to answer what a tool does, when to use it, what inputs it accepts, and what it returns, with concrete format examples. - Consolidation and Architectural Reduction: Reduce overlapping tool sets into comprehensive or primitive general-purpose tools, backed by production evidence such as the Vercel case reducing 17 tools to 2. - Schema Builder and Evaluator: Use the included Python utilities to build tool schemas, render markdown descriptions, score description quality, and generate agent-recoverable error messages. - Use Case: When implementing MCP tools for a multi-server agent, apply the fully qualified ServerName:tool_name naming convention and audit descriptions against the five evaluation criteria before deployment. ## Quick Start Ask the agent to design a new tool schema for your agent system, including description, parameters, return format, and error handling, following the tool-design guidelines.

Frequently Asked Questions about tool-design

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a good tool description for an AI agent?

Structure the description to answer four questions: what the tool does, when to use it, what inputs it accepts, and what it returns. Include concrete format examples like 'CUST-######' and document error conditions with recovery guidance.

How many tools should an agent have?

Limit collections to 10-20 tools for most applications, since description overlap causes model confusion. When more tools are needed, use namespacing such as db_* or web_* prefixes to create logical groupings that aid selection.

When should I consolidate multiple agent tools into one?

Consolidate when tools represent sequential steps in a single workflow, such as merging list_users, list_events, and create_event into one schedule_event tool. Keep tools separate when they have fundamentally different behaviors or serve different contexts.

How do I name MCP tools to avoid tool not found errors?

Always use fully qualified names in the format ServerName:tool_name, such as BigQuery:bigquery_schema. Without the server prefix, agents may fail to locate tools when multiple MCP servers expose similarly named tools.

Why do agents fail to use tools correctly?

Failures typically stem from vague descriptions, cryptic parameter names, missing usage context, and error messages without recovery guidance. Feed observed failures back to an agent to diagnose issues and improve descriptions iteratively.

When should I avoid reducing my agent's tool architecture?

Avoid architectural reduction when the data layer is messy or undocumented, the domain requires specialized knowledge the model lacks, safety constraints must limit agent actions, or workflows genuinely benefit from structured orchestration.