AI Tools Toolkit

AI Tools

MCP config explorer, system prompt builder with token counter, and agent skills reference for Claude and LLM development.

MCP Config JSON (claude_desktop_config.json or settings.json)

Servers

Env Vars

⚠️ Config contains placeholder secrets — replace before use

🔌

filesystem

Read/write local files and directories

File System

Commandnpx -y @modelcontextprotocol/server-filesystem /tmp

🔌

github

Access GitHub repos, issues, PRs, and code

Version Control

Commandnpx -y @modelcontextprotocol/server-github

Environment Variables (1)

GITHUB_PERSONAL_ACCESS_TOKEN=<your-token>

🔌

postgres

Query PostgreSQL databases with SQL

Database

Commandnpx -y @modelcontextprotocol/server-postgres postgresql://localhost/mydb

Known MCP Servers

filesystem

Read/write local files and directories

github

Access GitHub repos, issues, PRs, and code

postgres

Query PostgreSQL databases with SQL

sqlite

Query SQLite databases

brave-search

Web search using Brave Search API

slack

Read and post Slack messages

memory

Persistent key-value memory across sessions

puppeteer

Browser automation and web scraping

#mcp#system prompt#agent#claude#llm#ai#token

What's included in AI Tools Toolkit

3 tools, all free and client-side — no login or data upload required.

MCP Config Explorer

Validate and explore Model Context Protocol (MCP) server configuration JSON — check server names, commands, and arguments.

Validate MCP server configs before adding to Claude Desktop
Audit MCP tool names and argument schemas
Debug MCP configuration errors in AI agent setups

System Prompt Builder

Compose structured system prompts for Claude, GPT, and other LLMs with live token counting and role section templates.

Design role-specific personas for AI assistants
Optimize system prompts to fit within context window limits
Build and iterate on structured prompts with token budget awareness

Agent Skills Reference

Searchable reference for Claude agent skills, hooks, slash commands, and MCP integration patterns.

Look up available Claude Code slash commands and their behavior
Find hook event types for pre/post tool-call automation
Discover MCP patterns for building custom AI agent tools

About AI Tools Toolkit

The AI Tools Toolkit is built for developers working with large language models and AI agents — particularly those building on Claude, the Model Context Protocol (MCP), and agent frameworks. Validate MCP server configuration JSON before adding it to Claude Desktop, compose structured system prompts with live token counting to stay within context window limits, and search a reference of Claude agent skills, hooks, and slash commands.

The System Prompt Builder supports templates for common roles (assistant, code reviewer, data analyst) and breaks prompts into sections (role, instructions, examples, constraints) with token counts per section — so you can see exactly where your budget is going and iterate on the parts that matter most.

Like the rest of the site, every tool runs locally. No prompts, configs, or agent definitions are sent to any model or logged anywhere. This makes the toolkit safe for developing proprietary agents, internal tooling, and confidential prompt templates.

How to use AI Tools Toolkit

1
Pick an AI tool
Choose a tab: MCP Config Explorer, System Prompt Builder, or Agent Skills Reference.
2
Paste or compose
For MCP, paste your server config JSON. For System Prompt Builder, fill the role, instructions, and examples sections. For Agent Skills, search by keyword.
3
Validate and iterate
MCP config is validated against the protocol schema. System prompts show token counts live. Agent skills shows descriptions and usage patterns.
4
Copy the output
Copy validated MCP JSON, a finalized system prompt, or a skill reference into your project.

Frequently asked questions

What is the Model Context Protocol (MCP)?

MCP is an open protocol developed by Anthropic that lets AI applications (like Claude Desktop) connect to external tools and data sources through a standardized server interface. MCP servers expose tools, resources, and prompts that the AI can invoke to accomplish tasks.

Does the System Prompt Builder use the official Claude tokenizer?

It uses a close approximation (based on the public tokenizer spec) that is accurate within ~2% for English and well within the margin needed for context-window budgeting. For final production counts, verify with the API's /messages/count_tokens endpoint.

Can I export a System Prompt Builder prompt as Claude API JSON?

Yes. The exported format can be pasted directly as the 'system' parameter of a Claude API message request, or inserted into any OpenAI-compatible system message field.

What does the Agent Skills Reference cover?

It covers Claude Code's skill system: user-invocable skills, auto-triggered skills, hook events (PreToolUse, PostToolUse), slash commands, and MCP integration patterns. Each entry includes a description and when to use it.

Is the MCP Config Explorer up to date with the latest protocol version?

Yes. It validates against the current MCP specification including tool, resource, and prompt primitives, plus the server capabilities declaration.

Can these tools help me build an agent for my company?

Yes. The System Prompt Builder helps define role and constraints, MCP Explorer helps wire up external data sources, and the Skills Reference shows what Claude can be extended to do. All three are designed for production agent development.