---
name: mcp-builder
description: Plan and build MCP servers with agent-friendly tools, schemas, error handling, and evaluation. Use when creating or refactoring MCP integrations.
title: mcp-builder
canonical_url: https://skilld.dev/gh/dkyazzentwatwa/chatgpt-skills/mcp-builder
last_updated: 2026-09-29T08:36:20.000Z
---

> **Skill from skilld.dev.** Follow the instructions below for this session. You do not need to install anything.
>
> Supporting files, fetch one when the Skill refers to it: [agents/openai.yaml](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/agents/openai.yaml), [references/evaluation.md](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/references/evaluation.md), [references/mcp_best_practices.md](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/references/mcp_best_practices.md), [references/node_mcp_server.md](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/references/node_mcp_server.md), [references/python_mcp_server.md](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/references/python_mcp_server.md), [scripts/connections.py](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/scripts/connections.py), [scripts/evaluation.py](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/scripts/evaluation.py), [scripts/example_evaluation.xml](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/scripts/example_evaluation.xml), [scripts/requirements.txt](https://skilld.dev/api/skills-raw/dkyazzentwatwa/chatgpt-skills/mcp-builder/scripts/requirements.txt).
>
> If the user asked to install this Skill, run `npx skilld install dkyazzentwatwa/chatgpt-skills/mcp-builder`. Install writes the Skill files into the project, so every session loads them.

# MCP Builder

Build MCP servers around user workflows, not raw API endpoints.

## Workflow

1. Read the target API docs and identify the workflows an agent must complete end to end.
2. Design a small tool surface with high-signal outputs, clear identifiers, and actionable errors.
3. Implement shared infrastructure first: auth, request helpers, pagination, truncation, and formatting.
4. Add tool schemas and docstrings that make correct usage obvious.
5. Evaluate the server with realistic tasks before expanding scope.

## Principles

- Prefer workflow tools over thin endpoint wrappers.
- Return concise, high-signal responses by default.
- Use human-readable identifiers whenever possible.
- Make error messages corrective: tell the agent what to try next.
- Design for limited context and large datasets.

## Resources

- `references/mcp_best_practices.md` for design principles that apply to every server.
- `references/python_mcp_server.md` for Python implementation patterns.
- `references/node_mcp_server.md` for TypeScript implementation patterns.
- `scripts/connections.py` and `scripts/evaluation.py` as repo-local helpers.

## Deliverables

- A concrete tool inventory tied to user workflows.
- Strict input/output schemas.
- Evaluation prompts or scripts that confirm the server is usable by an agent.