discovering-tools

Discover MCP servers and tools with just-in-time schema loading.

19|3|Updated Dec 15, 2025

Search Tags:#workflow #MCP #discovery #tools #schemas #progressive-disclosure #just-in-time

Ready to run

npx skills add https://github.com/GobbyAI/gobby --skill discovering-tools

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: discovering-tools
Source: https://github.com/GobbyAI/gobby/tree/main/src/gobby/install/shared/skills/discovering-tools
Command: npx skills add https://github.com/GobbyAI/gobby --skill discovering-tools

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill provides a progressive discovery pattern for MCP tools and skills, avoiding the need to preload all servers, tools, and schemas.

Core Features & Use Cases

Lightweight discovery: enumerate MCP servers and tools without fetching full schemas.
Just-in-time schema loading: retrieve full tool schemas only when they are actually needed.
Skill discovery parity: apply the same progressive approach to skills for responsive UX in toolchains.

Quick Start

Use a just-in-time discovery flow: list_mcp_servers(), list_tools() for a server, get_tool_schema() on demand, and call_tool() after validating input.

Frequently Asked Questions about discovering-tools

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

FAQPage Schema

How do I reduce token usage when discovering MCP tools across multiple servers?▼

Progressive discovery for MCP tools minimizes token usage and latency by loading lightweight metadata first, then fetching full schemas only on demand. You can enumerate servers and tools without preloading all schemas.

What is just-in-time schema loading for MCP environments?▼

Just-in-time schema loading is a progressive discovery pattern that retrieves full MCP tool schemas only when needed. It applies a stepwise flow: list servers, list tools, get schemas on demand, and call tools with validated parameters.

How do I list MCP servers and tools without fetching full schemas?▼

You can list MCP servers and tools without fetching full schemas by using lightweight discovery functions like list_mcp_servers() and list_tools(). This avoids preloading all servers and tools, minimizing latency.

Can I apply progressive discovery to skills as well as MCP tools?▼

Yes, skill discovery parity applies the same progressive approach to skills. This ensures responsive UX in toolchains by using just-in-time discovery to load metadata first and full schemas on demand.

What is the stepwise flow for calling MCP tools with validated parameters?▼

The stepwise flow to call MCP tools involves listing servers, listing tools, getting full tool schemas on demand, and calling tools with validated parameters. This just-in-time discovery pattern minimizes token usage.

Do I need to preload all MCP servers and tools before using them?▼

No, you do not need to preload all MCP servers and tools. Progressive discovery avoids preloading by fetching lightweight metadata first and retrieving full schemas only when they are actually needed.