VividKit Crystal clear AI coding
VI
Choose Your Setup Get Started with AgentKit Continue with ClaudeKit Migrate from ClaudeKit to AgentKit Clean cutover from ClaudeKit What Is AgentKit? How AgentKit Works AgentKit Kits, Skills, Agents & Hooks AgentKit Runtime Support AgentKit Desktop App AgentKit Configuration AgentKit CLI Quick Reference AgentKit Skills Cheatsheet AgentKit Workflows AgentKit Statusline Update AgentKit CLI & Kits AgentKit Helper Clean reinstall of AgentKit AgentKit Troubleshooting What is ClaudeKit? ClaudeKit CLI Guide Command Finder Commands Cheat Sheet Workflow Recipes CLI Quick Reference Claude Mechanics Permission Modes ClaudeKit Hooks Coexistence ClaudeKit x Codex CK → Other coding tools How ClaudeKit Works Inside ClaudeKit Finding Your Unknowns Session Recovery Fix from Logs UI/UX Guide Accounts Switcher IDE Config Remote Control Codex App System Diagramming UI Review Gate Promotions Support VividKit

Quick Ref / Command

Marketing Kit

/ak:use-mcp

Run existing MCP tools

Discover and execute MCP server tools through the narrowest safe path: runtime-native tools first, deterministic direct scripts when needed.

01

Choose path

02

Discover native

03

Select minimal

04

Fallback scripts

Rule 01

Use runtime-native MCP when visible; it has the lowest setup and context cost.

Rule 02

Use direct scripts for deterministic tool names, schemas, and servers declared outside the active runtime.

Rule 03

Capability discovery comes before execution.

Rule 04

MCP consumption is not MCP server creation; mcp-builder is not a fallback.

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Choose path Decide between runtime-native MCP and direct scripts based on whether the active runtime already exposes the needed server or tool.
  2. 2 Discover native Inspect resources and tools visible in Claude Code, Codex, or Pi; use actual names instead of assuming shared spelling.
02

Work

Agent / skill execution

  1. 3 Select minimal Pick the smallest tool that directly satisfies the task and validate required arguments before execution.
  2. 4 Fallback scripts If the server exists only in .claude/.mcp.json or needs deterministic schemas, install scripts dependencies and use cli.ts list/call commands.
03

Verify

Gate, review, validation

  1. 5 Catalog Run list-tools to persist assets/tools.json with schemas, then review it before calling unknown tools.
  2. 6 Execute Call the selected MCP tool with validated JSON arguments, timeout handling, and lifecycle cleanup from the script client when applicable.
04

Close

Report, handoff, artifact

  1. 7 Handle limits For Chrome cookie/profile work invoke ak:chrome-profile first; do not use mcp-builder as a fallback for consuming existing servers.
  2. 8 Report Explain which path was used, what capability was found, what was executed, and any server visibility caveat.

Special Operations

Runtime-native first

If the tool is already exposed by the active session, avoid extra client setup and call the visible capability directly.

Direct scripts

Use the bundled client for servers only declared in .claude/.mcp.json or for reproducible scripted workflows.

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Tool discovery Recommended 
/ak:use-mcp discover the available browser automation tools
Use when:
Use when you need to know what MCP capabilities are available before choosing one.
Expected:
A minimal catalog of relevant tools, argument requirements, and safe execution boundaries for the requested integration.
Controlled execution 
/ak:use-mcp call the notebook source list tool for this project
Use when:
Use when the task requires executing a specific MCP capability safely.
Expected:
Validated MCP call path with the observed result or a clear visibility caveat when unavailable.
Project MCP catalog 
/ak:use-mcp discover tools from .claude/.mcp.json for this project
Use when:
The runtime does not expose a server but the project MCP config exists.
Expected:
Uses the scripted catalog path against .claude/.mcp.json without copying credentials across runtimes.

Handled Scope

  • MCP tool discovery
  • Schema-checked tool execution
  • Direct script client workflows
  • Persistent tool catalogs
  • Runtime visibility caveats

Related skills

Next