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

Engineer Kit

/ak:mcp-builder

High-signal MCP servers

Design and build high-signal MCP servers around real agent workflows, not thin endpoint wrappers.

01

Study agent design

02

Load MCP references

03

Study the API

04

Plan tools and contracts

Rule 01

Build workflow tools, not endpoint wrappers

Rule 02

Optimize every response for scarce agent context

Rule 03

Errors must teach the next action

Rule 04

Evaluate with realistic tasks before calling the server useful

Execution Map

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

01

Start

Input, scope, route

  1. 1 Study agent design Start from workflow-first tools, limited context, actionable errors, natural task grouping, and evaluation-driven iteration.
  2. 2 Load MCP references Fetch the current MCP protocol, best-practices reference, and the Python or TypeScript SDK material for the chosen stack.
02

Work

Agent / skill execution

  1. 3 Study the API Read the service documentation for auth, rate limits, pagination, error shapes, endpoints, parameters, and data models.
  2. 4 Plan tools and contracts Choose high-value workflow operations, shared helpers, validation schemas, response formats, truncation limits, and error handling.
03

Verify

Gate, review, validation

  1. 5 Set up the server Create the Python/FastMCP or Node/TypeScript project shape with Pydantic or Zod validation and SDK registration patterns.
  2. 6 Implement shared infrastructure Build API request helpers, auth/token handling, pagination utilities, response formatting, and reusable external-call errors first.
04

Close

Report, handoff, artifact

  1. 7 Register and review tools Add each tool with strict inputs, descriptions, async I/O, annotations, language checklist review, and safe build/syntax checks.
  2. 8 Create evaluations Load the evaluation guide, inspect tools, explore read-only data, create 10 realistic verified questions, and output XML.

Syntax, arguments, subcommands

How to invoke the skill: syntax, positional arguments, shared options, then each subcommand with its own syntax and outcome.

Syntax

/ak:mcp-builder [service or API to integrate]

Arguments

[service or API to integrate] Required

Target integration

External service or API to turn into a new MCP server or expanded tool/resource surface. Include intended workflows, language, transport, auth model, permissions, rate limits, data volume, and allowed external effects when known.

/ak:mcp-builder "Build a TypeScript MCP server for the Acme tickets API with read-only search first, cursor pagination, and stdio transport"

Sample Prompt

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

Default API server Recommended 
/ak:mcp-builder Linear issue triage API
Use when:
Invoke when building an MCP server or tool surface for an external service.
Expected:
Researches MCP references and Linear API docs, plans workflow-level tools, implements shared helpers and schemas, then creates verified evaluation questions.
Python FastMCP server 
/ak:mcp-builder FastMCP server for customer search
Use when:
Use when the MCP surface should be built in Python with Pydantic validation.
Expected:
Loads Python MCP guidance, sets up FastMCP registration, defines Pydantic inputs, implements async API helpers, and verifies usefulness with read-only evaluations.
TypeScript MCP server 
/ak:mcp-builder TypeScript MCP server for billing admin workflows
Use when:
Use when the host project needs a Node/TypeScript MCP SDK server.
Expected:
Loads TypeScript SDK guidance, configures project structure and strict Zod schemas, registers tools with annotations, runs build checks, and prepares XML evaluations.
Evaluation-focused build 
/ak:mcp-builder MCP server for read-only knowledge base questions
Use when:
Use when tool usefulness must be proven against stable, realistic agent questions.
Expected:
Designs read-only workflows, inspects the implemented tools, explores available content, writes 10 independent verifiable questions, and emits the required XML format.

Handled Scope

  • FastMCP Python servers
  • TypeScript MCP SDK servers
  • Agent-centric tool design
  • Schema validation and annotations
  • Evaluation-driven server refinement

Artifacts Produced

MCP Deliverables

Pattern

Server code + tool schemas + evaluations

A task-oriented MCP server with concise responses, strict inputs, actionable errors, safe long-running test strategy, and realistic evaluation questions.

Next