Chat Thing's MCP server: build AI agents by describing them [Chat Thing](https://chatthing.ai/) Open main menu - Features - Use Cases - [Pricing](https://chatthing.ai/#pricing) - Resources [Log in](https://app.chatthing.ai/app/auth/login) [ Book a demo](https://chatthing.ai/demo) [Sign up](https://app.chatthing.ai/app/auth/register) [← Back to blog](https://chatthing.ai/blog)Blog # Chat Thing's MCP server is live: describe your agent, let AI build it ![Zef](https://res.cloudinary.com/djyjvrw5u/image/upload/v1710941716/IMG_2278_d3b5e4fa69.jpg) Zef 15 Jul 2026 ~ 6 min read --- ![Chat Thing's MCP server is live: describe your agent, let AI build it](https://res.cloudinary.com/djyjvrw5u/image/upload/v1784134253/cover_bd9b756581.png) --- ### On this page 1. [Connect in a minute](#connect-in-a-minute) 2. [Describe it, don't configure it](#describe-it-dont-configure-it) 3. [It tests and fixes its own work](#it-tests-and-fixes-its-own-work) 4. [Custom integrations, configured and tested for you](#custom-integrations-configured-and-tested-for-you) 5. [It looks after what's already live](#it-looks-after-whats-already-live) 6. [Things to ask it](#things-to-ask-it) 7. [Get started](#get-started) We've spent a long time making Chat Thing simple. It's still a semi-technical product. To get a really good agent out of it, you have to know things. Which model suits the job. What enhanced retrieval does, and why a follow-up question falls apart without it. Why your FAQ entries aren't surfacing (the relevance threshold is too high for short rows). How to write a prompt that stays on topic without going cold. How to shape a power-up's config so it actually fires. Most people who need a support agent don't want to learn any of that. They want an agent that answers questions properly. So we've handed the technical decisions to something that already knows them: the AI you're already talking to. Chat Thing now has an MCP server, and you describe what you want in plain English while it does the rest. This is not a developer feature. If you use the Claude app or ChatGPT, you can connect Chat Thing in a couple of clicks and build an agent by asking for one. No terminal, no config file, no code. ## Connect in a minute Everything points at the same endpoint: ``` https://app.chatthing.ai/mcp ``` **If you use the Claude app or ChatGPT**, paste that URL into the connectors panel. In Claude (desktop or claude.ai) it's Settings, then Connectors, then Add custom connector. In ChatGPT you'll need to turn on developer mode under Settings, Apps & Connectors, Advanced settings first, then create the connector. Leave the advanced OAuth fields blank in both. We register your client automatically. **If you live in an editor**, Claude Code is one line: ``` claude mcp add chatthing --transport http https://app.chatthing.ai/mcp ``` Cursor, Windsurf, VS Code, Zed, Cline and the Codex CLI take a few lines of JSON at the same URL. Anything that speaks streamable HTTP works, listed or not. Either way there's no key to paste. Your browser opens, you approve the team, you're connected. Or hand your AI this and let it read the manual itself: > Set up ChatThing for me. Fetch https://chatthing.ai/docs/mcp-getting-started.md and follow the instructions. ## Describe it, don't configure it One sentence: > Build me a customer support agent. Use the help pages at https://help.example.com as its knowledge, and add a way for customers to reach a human if it can't answer. That's the technical brief now. Your agent works out the rest: it previews which pages it would index so you can argue with the list before anything gets crawled, adds them, picks a model, turns on enhanced retrieval because it's a multi-page support agent that'll get follow-up questions, and wires up the escalation power-up with the right fields. You didn't have to know what enhanced retrieval was. It did. ![One sentence in the Claude app, and the whole build runs: Claude checks which teams you own, picks a model, discovers 114 pages on pixelhop.io, skips the 36 tag and pagination pages because they are just link lists, creates the agent, syncs 78 pages and attaches the talkToAHuman hand-off.](https://res.cloudinary.com/djyjvrw5u/image/upload/v1784134207/hero_1594fd4811.png) ## It tests and fixes its own work This is the part that changes the job. ![Claude testing the new agent and catching it making something up: it spots the agent describing Rye & Beyond as "an AI product from prototype to launchable in weeks" when it is actually a JAMStack holiday-rental site from 2020, then checks a suspicious founding-date answer against the real about page and finds the site itself says both 2017 and 2019.](https://res.cloudinary.com/djyjvrw5u/image/upload/v1784134212/testing_4f4bfdc27d.png) Your agent doesn't hand you an agent and wish you luck. It can hold a real conversation with what it just built, ask questions it knows the answers to, and read back which sources got cited. When the answer's wrong, it can see _why_ it's wrong: the wrong page cited, nothing retrieved, the prompt drifting off topic. Then it fixes it and asks again. Which means you can debug in plain English too: > It keeps saying it doesn't know our refund policy, but that's on the returns page. And it'll go and check whether the page got indexed, whether the retrieval settings are letting it through, and whether the prompt is getting in the way. That loop used to be you, clicking between four screens with a hunch. On Enterprise, it can also build proper eval suites: write test cases, run them, read the results, so you catch a regression before your customers do. ## Custom integrations, configured and tested for you Power-ups are where your agent stops answering questions and starts doing things: looking up an order, updating a record, triggering a workflow. The API power-up is the most useful of them and by far the most technical. You're describing an endpoint, its auth, and a schema that tells the model when to call it and what to send. Get a field wrong and your agent either never calls it or calls it with rubbish. It's the one place in Chat Thing where no-code runs out of road for most people. It's also exactly what a coding agent is good at. It knows JSON schemas, it knows how your auth works, and if it has your codebase open it knows your API better than you'd manage to describe it from memory. ![Claude configuring the API power-up: it fetches the useAPI schema first, wires it up as a POST with a structured argument set so the agent collects real project details rather than dumping a transcript, then tests it end to end with a realistic enquiry and checks whether unfilled placeholders leak into the payload.](https://res.cloudinary.com/djyjvrw5u/image/upload/v1784134217/configure_powerup_1f5f68ac92.png) Then it proves it. It sets the power-up up, asks your agent something that should trigger the call, and looks at what actually happened. If the config is wrong, Chat Thing hands back per-field errors with the schema attached, so it corrects itself instead of guessing at your field names. Fix, ask again, repeat until the thing genuinely fires. A custom integration, built by describing it, and working in front of you before you ship it. ## It looks after what's already live Your agents don't stop needing attention once they're deployed. Your AI can read what's actually happening in them: > Summarise the last week of conversations with my support agent. What are people asking that it's struggling with? It reads the real chats, spots the patterns, and tells you where the gaps are. Then it can close them: add the missing page, tune the prompt, add a power-up for the thing everyone keeps asking to do. The reporting and the fix are the same conversation. That one's worth doing from the Claude app on a Monday morning, coffee in hand. It's the sort of thing that used to mean a data export and a spreadsheet. ## Things to ask it It reaches **47 tools**, very nearly the whole product. Some of what that means in practice: > Train it on our YouTube channel and blog feed as well as the help site. Web pages, RSS and YouTube, without you working out which is which. > Ping our Slack whenever someone asks to speak to a human. Webhooks, with a Slack preset, so you don't hand-write the JSON. > We've rewritten our returns policy. Re-sync the site and check the agent gives the new answer. Re-crawls, then proves it worked by asking. > Make the widget match our brand colours and change the welcome message. Your website widget, styled without touching CSS. > Swap this agent onto a faster model and check the answers still hold up. Changes the model, re-runs your questions, tells you what moved. > Build a second agent for our docs site, same setup as the support one. The bit that used to be an afternoon of clicking. ## Get started Connect your client in about a minute: [chatthing.ai/mcp](https://chatthing.ai/mcp), or read the [getting-started guide](https://chatthing.ai/docs/mcp-getting-started). Then just ask it to build you something. ### Related articles [![How ready is your website for AI agents? Now there's a free way to check](https://res.cloudinary.com/djyjvrw5u/image/upload/v1783957835/hero_91cf981c76.png)](https://chatthing.ai/blog/agent-readiness-checker) [

How ready is your website for AI agents? Now there's a free way to check

](https://chatthing.ai/blog/agent-readiness-checker) We built the Agent Readiness Checker: a free, no-hype audit of how well AI agents and LLMs can discover, understand and act on your site. And you can chat to an agent about the results. [![Connect Your AI Agent to 8,000+ Apps with Zapier](https://res.cloudinary.com/djyjvrw5u/image/upload/v1782482565/zapier_hero_debe7969e3.jpg)](https://chatthing.ai/blog/zapier) [

Connect Your AI Agent to 8,000+ Apps with Zapier

](https://chatthing.ai/blog/zapier) Connect your Chat Thing agent to 8,000+ apps with Zapier. Book meetings, update CRMs, post to Slack, all in one conversation. [![The work you don't see (and why it makes Chat Thing better)](https://res.cloudinary.com/djyjvrw5u/image/upload/v1780301027/og_A_engine_room_b535f0b904.png)](https://chatthing.ai/blog/the-work-you-dont-see-and-why-it-makes-chat-thing-better) [

The work you don't see (and why it makes Chat Thing better)

](https://chatthing.ai/blog/the-work-you-dont-see-and-why-it-makes-chat-thing-better) This month we rebuilt the engine room of Chat Thing. Not the bit you log into, but the part doing all the heavy lifting behind the scenes.. crawling your website, syncing your content from Notion, Google Drive and the rest, generating your analytics, sending your emails, running your agents' tasks. All of that runs on a separate system, and we tore it down and rebuilt it from scratch.
Previous
[← How ready is your website for AI agents? Now there's a free way to check](https://chatthing.ai/blog/agent-readiness-checker)
## Footer ### Ready to answer every question, day or night? Add an AI support agent to your site in minutes. [Get started free](https://app.chatthing.ai/app/auth/register) [Book a demo](https://chatthing.ai/demo) AI customer support that learns from your content and answers 24/7. ### Features - [Channels: Deploy your bot](https://chatthing.ai/pages/features/channels) - [Data sources: Add knowledge](https://chatthing.ai/pages/features/data-sources) - [Power-ups: Tool calling](https://chatthing.ai/pages/features/power-ups) - [AI Models: GPT-4, Claude, & more](https://chatthing.ai/pages/features/models) - [SDK: Create custom assistants](https://chatthing.ai/pages/features/chatbot-sdk) - [Bot evaluations: Test your bot](https://chatthing.ai/pages/features/testing) - [Analytics: Topics, sentiment & insights](https://chatthing.ai/pages/features/analytics) - [Bot Gallery: Example bots](https://chatthing.ai/gallery) - [All features](https://chatthing.ai/features) ### Use cases - [AI Customer Support Bots](https://chatthing.ai/pages/use-cases/customer-support) - [AI Internal Knowledge Bots](https://chatthing.ai/pages/use-cases/internal-knowledge) - [AI Bots for Marketing](https://chatthing.ai/pages/use-cases/marketing) - [SDK: Create custom assistants](https://chatthing.ai/pages/features/chatbot-sdk) - [All features](https://chatthing.ai/features) ### Channels - [AI support for your website](https://chatthing.ai/pages/features/channels/website) - [AI support for Slack](https://chatthing.ai/pages/features/channels/slack) - [AI support for Discord](https://chatthing.ai/pages/features/channels/discord) - [AI support for WhatsApp](https://chatthing.ai/pages/features/channels/whatsapp) - [AI support for Telegram](https://chatthing.ai/pages/features/channels/telegram) - [AI support API](https://chatthing.ai/pages/features/channels/api) ### Data sources - [Chat with your website](https://chatthing.ai/pages/data-source/web) - [Chat with your files](https://chatthing.ai/pages/data-source/files) - [Notion chatbots](https://chatthing.ai/pages/data-source/notion) - [Chat with your youtube](https://chatthing.ai/pages/data-source/youtube) - [Chat with your RSS feed](https://chatthing.ai/pages/data-source/rss) ### Company - [Discord](https://discord.com/invite/DGCCR34Fsj) - [Docs](https://chatthing.ai/docs/getting-started) - [Blog](https://chatthing.ai/blog) - [FAQ](https://chatthing.ai/faq) - [Status](https://status.chatthing.ai/) - [Testimonials](https://senja.io/p/chat-thing/iq1tWz) - [Custom AI development](https://chatthing.ai/pages/ai-development-services) ### Free tools - [Agent Readiness Checker](https://chatthing.ai/tools/agent-readiness-checker) - [JSON-LD Generator + Checker](https://chatthing.ai/tools/json-ld-generator) - [llms.txt Generator & Validator](https://chatthing.ai/tools/llms-txt-generator) - [Support Ticket Deflection Calculator](https://chatthing.ai/tools/ticket-deflection-calculator) - [Knowledge Base Grader](https://chatthing.ai/tools/knowledge-base-grader) - [AI FAQ Generator](https://chatthing.ai/tools/faq-generator) - [AI Canned Response Generator](https://chatthing.ai/tools/canned-response-generator) - [Support Bot Prompt Generator](https://chatthing.ai/tools/support-bot-prompt-generator) - [All free tools](https://chatthing.ai/tools) ### Compare - [Intercom alternative](https://chatthing.ai/pages/alternatives/intercom) - [Zendesk alternative](https://chatthing.ai/pages/alternatives/zendesk) - [Tidio alternative](https://chatthing.ai/pages/alternatives/tidio) - [Crisp alternative](https://chatthing.ai/pages/alternatives/crisp) - [Chatbase alternative](https://chatthing.ai/pages/alternatives/chatbase) - [DocsBot alternative](https://chatthing.ai/pages/alternatives/docsbot) - [ResolveAI alternative](https://chatthing.ai/pages/alternatives/resolveai) - [ChatBotKit alternative](https://chatthing.ai/pages/alternatives/chatbotkit) - [CustomGPT alternative](https://chatthing.ai/pages/alternatives/customgpt) - [Botsonic alternative](https://chatthing.ai/pages/alternatives/botsonic) - [Ada alternative](https://chatthing.ai/pages/alternatives/ada) ### Join our newsletter Be the first to know about new features and updates. Join [Privacy](https://chatthing.ai/legal/privacy) [Terms](https://chatthing.ai/legal/terms) [DPA](https://chatthing.ai/legal/dpa) [GDPR](https://chatthing.ai/legal/gdpr) [Sub-Processors](https://chatthing.ai/legal/sub-processors) [Twitter](https://twitter.com/pixelhopio) [Discord](https://discord.com/invite/DGCCR34Fsj) [YouTube](https://www.youtube.com/@ChatThingAI) © 2026 Chat Thing. All rights reserved. Company number: 14789358. Built with by [Pixelhop](https://pixelhop.io?utm_source=chat+thing&utm_medium=website) ## Cookies We use cookies to run the site and understand how it's used. You can change your choice anytime. Accept Decline Customise