Model Context Protocol

The documentation, as a tool your agent can call

Your agent already guesses at Cursor flags, Claude Code hooks and Codex config. Point it here instead and it reads the guide — mid-prompt, without you leaving the editor.

Endpoint https://developertoolkit.ai/mcp
  • No account
  • No API key
  • Read-only

Setup

One line, then it is there

A remote HTTP server, so there is nothing to install and nothing to keep running. Add it once and both tools show up in your next session.

Claude Code

Terminal
claude mcp add --transport http dtk https://developertoolkit.ai/mcp

Cursor

.cursor/mcp.json
{
  "mcpServers": {
    "dtk": {
      "type": "http",
      "url": "https://developertoolkit.ai/mcp"
    }
  }
}

Codex

~/.codex/config.toml
[mcp_servers.dtk]
url = "https://developertoolkit.ai/mcp"

Any MCP client that speaks Streamable HTTP works — including ChatGPT connectors, where you paste the endpoint URL directly.

Try it

See what your agent gets back

This box calls the same endpoint your editor would, from your browser. The results below are the real response — no mock, no cached sample.

Try

Contract

Two tools, no surprises

The names are the pair ChatGPT and Claude connectors look for, so citations resolve on their own. Each result carries the id you pass straight to the second tool.

search

in
{ query: string }
out
{ results: [{ id, title, url, snippet, lang }] }

Ranked across both languages at once. A Polish query returns Polish articles without a language parameter — rare words carry more weight than common ones, and that alone separates the two halves of the corpus.

fetch

in
{ id: string }
out
{ id, title, text, url, metadata }

The complete article as markdown, not an extract. `metadata` reports the language, the section, the last-updated date, and whether the page sits behind the subscription on the site.

Corpus

What it can actually read

Every guide on this site, in both languages, rebuilt on each deploy. Cursor, Claude Code and OpenAI Codex, from first setup through hooks, MCP, testing, CI and deployment.

950+
articles indexed
2
languages, EN and PL
12
documentation sections
0
API keys required

Limits

What it does not do

It only reads

Two tools, both read-only. Nothing here can write to your repository, run a command, or reach anything but this site’s own documentation.

It keeps no session

Every call is independent and nothing about you is stored between them. That is also why the server needs no account and no key.

It is not the only way in

The same text is published at /llms.txt and /llms-full.txt for crawlers. MCP exists because downloading six megabytes to answer one question is not reading.