Skills mcp-builder

📦

mcp-builder

Name: mcp-builder
Author: 92Bilal26

Content revision r1 High Risk 🌐 Network access🔑 Env variables⚙️ External commands

Build Reliable MCP Servers

Building MCP servers can be slow because protocol details, schemas, transports, and evaluation design must align. This skill provides structured guidance, examples, and an evaluation harness for Python and TypeScript MCP integrations.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Stop for confirmation before installing.

Review the plan and obtain explicit user consent before changing files.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "mcp-builder" from https://skillstore.io/skills/92bilal26-mcp-builder.md and its manifest at https://skillstore.io/api/skills/92bilal26-mcp-builder/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/92bilal26-mcp-builder.md Signed manifest GET /api/skills/92bilal26-mcp-builder/manifest Signed lockfile GET /api/skills/92bilal26-mcp-builder/lockfile

Test it

Using "mcp-builder". Plan a GitHub issue tracking MCP server.

Expected outcome:

A phased implementation plan with tool names, required inputs, read-only annotations, pagination behavior, and evaluation ideas.

Using "mcp-builder". Review a Python MCP server design for a CRM API.

Expected outcome:

A checklist covering Pydantic models, async HTTP usage, error messages, response formats, and missing security controls.

Using "mcp-builder". Create evaluation questions for a support ticket MCP server.

Expected outcome:

Three stable read-only tasks with expected answer styles and notes about avoiding destructive operations.

Security Audit

High Risk

v7 • 7/5/2026 Open versioned report

Most static findings are false positives caused by Markdown examples, TypeScript template literals, placeholder URLs, and dummy credential names in reference documentation. No prompt-injection attempt or embedded real secret was found. The remaining concerns are operational: the evaluation harness can launch user-supplied commands, pass environment values, connect to arbitrary MCP URLs, and send tool outputs to Anthropic.

Files scanned

3,532

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (3)

High

scripts/evaluation.py:327-329 scripts/connections.py:76-84

Evaluation Harness Launches User-Supplied Commands

The evaluation script accepts a command and arguments, then the connection layer passes them to StdioServerParameters. Running untrusted evaluation settings could execute arbitrary local processes with supplied environment values.

The CLI exposes command, args, and env options, and the connection code uses them to create a stdio MCP subprocess. This is direct local process launch behavior, although it is intended for trusted evaluations.

Medium

scripts/evaluation.py:17 scripts/evaluation.py:96-103 scripts/evaluation.py:116-120 scripts/evaluation.py:137-144

Evaluation Data Sent To External LLM API

The evaluation harness sends questions, tool schemas, conversation history, and MCP tool results to Anthropic messages calls. Sensitive data returned by evaluated MCP tools could leave the local environment.

The script constructs Anthropic client calls with the evaluation messages and tools, and appends tool results into later requests. This clearly sends evaluation context to an external model API.

Medium

scripts/evaluation.py:331-333 scripts/connections.py:91-109

Remote MCP Connections Accept Arbitrary URLs And Headers

The harness accepts user-provided SSE or HTTP URLs and headers, then connects using MCP remote transports. Tokens in headers could be sent to an unintended endpoint if copied from untrusted instructions.

The command-line interface accepts arbitrary remote URLs and headers, and the connection layer passes them directly to SSE or streamable HTTP clients. The risk depends on user-supplied configuration.

Risk Factors

🌐 Network access (33)

🔑 Env variables (18)

⚙️ External commands (46)

Audited by: codex View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/92bilal26-mcp-builder/audits/7?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/92bilal26-mcp-builder/security.svg)](https://skillstore.io/skills/92bilal26-mcp-builder?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/92bilal26-mcp-builder?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/92bilal26-mcp-builder/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/92bilal26-mcp-builder.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

92Bilal26. (2026). mcp-builder security audit report (audit version 7) [Author version unspecified]. Skillstore. https://skillstore.io/skills/92bilal26-mcp-builder/audits/7

BibTeX citation

@techreport{92bilal26-92bilal26-mcp-builder-2026,
  author = {92Bilal26},
  title = {mcp-builder security audit report (audit version 7)},
  institution = {Skillstore},
  year = {2026},
  number = {7},
  url = {https://skillstore.io/skills/92bilal26-mcp-builder/audits/7},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "mcp-builder security audit report (audit version 7)"
version: "unspecified"
type: report
authors:
  - name: "92Bilal26"
date-released: "2026-07-05"
url: "https://skillstore.io/skills/92bilal26-mcp-builder/audits/7"
identifiers:
  - type: other
    value: "skillstore:92bilal26-mcp-builder:audit:7"
    description: "Skillstore immutable audit report identifier"

Compare variants

15 installable variants

Each author remains a separate installable skill. The recommended variant is ranked by Skillstore evidence.

Why this variant is first

Highest Skillstore Score

Author / source Skillstore Score Evidence Confidence Skillstore usage Updated

Doyajin174 Recommended

doyajin174-mcp-builder

Skillstore Score 78

Evidence Confidence High

Skillstore usage 13

Updated

2026-08-21

anthropics

mcp-builder

Skillstore Score 75

Evidence Confidence High

Skillstore usage 15

Updated

2026-08-21

Azeem-2

azeem-2-mcp-builder

Skillstore Score 75

Evidence Confidence High

Skillstore usage 8

Updated

2026-08-21

mcp-use

mcp-use-mcp-builder

Skillstore Score 68

Evidence Confidence High

Skillstore usage 10

Updated

2026-08-21

shubhamsaboo

shubhamsaboo-mcp-builder

Skillstore Score 67

Evidence Confidence Medium

Skillstore usage 5

Updated

2026-08-21

YYH211

yyh211-mcp-builder

Skillstore Score 66

Evidence Confidence High

Skillstore usage 68

Updated

2026-08-21

Cam10001110101

cam10001110101-mcp-builder

Skillstore Score 66

Evidence Confidence High

Skillstore usage 10

Updated

2026-08-21

DYAI2025

dyai2025-mcp-builder

Skillstore Score 66

Evidence Confidence High

Skillstore usage 7

Updated

2026-08-21

ZhanlinCui

zhanlincui-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 22

Updated

2026-08-21

92Bilal26 Current

92bilal26-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 17

Updated

2026-08-21

sickn33

sickn33-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 13

Updated

2026-08-21

ArtemisAI

artemisai-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 10

Updated

2026-08-21

ComposioHQ

composiohq-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 9

Updated

2026-08-21

AutumnsGrove

autumnsgrove-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 9

Updated

2026-08-21

davila7

davila7-mcp-builder

Skillstore Score 38

Evidence Confidence Medium

Skillstore usage 7

Updated

2026-08-21

Skillstore Score

Why this score Evidence Confidence: Medium

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Plan a New MCP Integration

Define tools, schemas, transports, and response formats before writing a server.

Review Server Quality

Check an MCP implementation for naming, validation, pagination, error handling, and tool annotations.

Create Evaluation Tasks

Design read-only questions and run an evaluation harness to measure tool usefulness.

Try These Prompts

Plan a Starter Server

Use mcp-builder to plan a basic MCP server for [service]. Include transport choice, tool names, schemas, response formats, and first test steps.

Draft an Implementation

Use mcp-builder to draft a [TypeScript or Python] MCP server design for [service]. Focus on validation, pagination, error handling, and annotations.

Design Evaluations

Use mcp-builder to create read-only evaluation tasks for my MCP server. Make each question stable, realistic, and easy to verify.

Audit Production Readiness

Use mcp-builder to review my MCP server plan for production readiness. Check security, authentication, schemas, tool descriptions, transport choice, and evaluation coverage.

Best Practices

Read current MCP protocol and SDK documentation before implementing server details.
Use schema validation, pagination, clear annotations, and bounded responses for every tool.
Run builds and evaluations only with trusted commands, credentials, URLs, and test data.

Avoid

Copying placeholder URLs, tokens, or service names into production code.
Running evaluation commands from untrusted repositories or generated instructions.
Returning unbounded API responses without filters, limits, or truncation messages.