skills.rest
vTRKA
vTRKACOMMUNITY
Explore all skills by this author →
GitHub

mcp-discovery

Select MCP capabilities from live runtime registry evidence.

Updated Apr 27, 2026
Search Tags:#browser automation#figma#mcp#devtools#tool routing#fallback handling#runtime discovery
One-click install
npx skills add https://github.com/vTRKA/supervibe --skill mcp-discovery-vtrka
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mcp-discovery
Source: https://github.com/vTRKA/supervibe/tree/main/skills/mcp-discovery
Command: npx skills add https://github.com/vTRKA/supervibe --skill mcp-discovery-vtrka

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents wasted planning and incorrect tool claims by discovering which MCP servers and tool families are actually available at runtime on the current host before an agent tries browser, Playwright, DevTools, desktop, or design/crawl capabilities.

Core Features & Use Cases

  • Runtime MCP registry validation: Selects MCP capability based on the live registry evidence stored in .supervibe/memory/mcp-registry.json, avoiding assumptions from config-only intent.
  • Host-correct capability gating: Enforces fail-closed behavior when tools are missing, stale, host-mismatched, or not exposed by the current host tool palette.
  • Capability-specific evidence separation: Distinguishes browser vs playwright vs chrome-devtools vs figma so the agent only claims the proof surface the runtime registry supports.
  • Safe fallbacks with confidence caps: Routes to visible non-MCP fallbacks when runtime proof is unavailable and caps confidence when freshness is unknown or stale.

Quick Start

Ask your AI to run MCP discovery at session start and then route the work to the best available capability for browser, Playwright, DevTools, desktop, Figma, or current-docs based on .supervibe/memory/mcp-registry.json.

Frequently Asked Questions about mcp-discovery

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

FAQPage Schema
How do I prevent my AI agent from using browser or Playwright tools that aren't actually available at runtime?▼

Runtime MCP discovery prevents incorrect tool claims by validating available MCP servers from a live registry before an agent attempts browser, Playwright, or DevTools automation. It enforces fail-closed behavior when tools are missing or host-mismatched.

What is MCP tool routing and when do I need it for web crawling or Figma extraction?▼

MCP tool routing selects capabilities based on live runtime evidence rather than config assumptions. You need it during session start or when tasks involve interactive external capabilities like web crawling, Playwright automation, or Figma extraction to avoid wasted planning.

How do I safely route tasks to available MCP tools using a registry file?▼

Route tasks by selecting MCP capabilities from live runtime discovery evidence stored in `.supervibe/memory/mcp-registry.json`. This enforces fail-closed selection, host-correct tool palette checks, and explicit fallback routing when runtime tool lists are missing.

Can I use MCP discovery to distinguish between browser, Playwright, and Chrome DevTools capabilities?▼

Yes, MCP discovery provides capability-specific evidence separation that distinguishes browser, Playwright, Chrome DevTools, and Figma tools. This ensures the agent only claims the proof surface that the runtime registry actually supports for each distinct capability.

What happens when the MCP registry is stale or runtime tool lists are missing during automation tasks?▼

When runtime proof is unavailable, MCP discovery routes to visible non-MCP fallbacks and caps confidence when registry freshness is unknown or stale. This prevents overconfident tool usage during browser, DevTools, or desktop verification tasks.

Do I need a specific environment setup to run MCP discovery for desktop and design tooling?▼

No dependencies are required, but the current host must expose the intended tool palette. MCP discovery checks host-correct capabilities and enforces fail-closed gating if tools are missing, stale, or not exposed by the current host environment.

Related Skills

Explore More Software Engineering Skills→
aleph23aleph23

mcp-management

Discover, analyze, and execute MCP capabilities across multiple servers.

Community
Advanced
sc30gswsc30gsw
11

chrome

Orchestrate Chrome DevTools MCP for DOM inspection, network monitoring, and automated browser interactions.

Community
Advanced
buvisbuvis
2

audit-mcp-health

Cross-reference configured MCP servers against the live deferred tool list.

Official
Advanced
berkayoztuncberkayoztunc
21

MCP Server Card Support

Publish MCP server metadata at /.well-known/mcp/server-card.json for automated discovery.

Community
Basic