Direct answer
Add an MCP Server to Claude Code
To add an MCP server to Claude Code, open the MCP or connectors settings in Claude Code, or the MCP configuration it reads, and add a remote server with the URL https://zambo.dev/mcp. No API key is needed. Reload the session, confirm the Zambo tools are listed, and run one small call you can verify by eye. This page covers the setup, the default and full tool surfaces, and what the connection adds to real coding work.
Updated October 20, 2026.
What an MCP server adds to Claude Code
Claude Code already works in your terminal and your repo: it reads files, edits code, and runs commands. An MCP server extends what a session can reach beyond the repo. MCP, the Model Context Protocol, is a standard way for an AI client to plug in outside tools from a server at one address. Connect a server and the session gains those tools alongside its coding tools: live web and data checks when the code depends on the outside world, current facts instead of training data, memory that survives the session, and a record of the outside calls it made.
That record is the part developers underestimate until they need it. A Claude Code session acts with real latitude, and afterwards the summary it gives you is the agent grading its own homework. Tool calls made through Zambo, the verifiable work layer for AI, each return a verifiable receipt for the saved result, checkable by anyone free. The session can still be wrong, and receipts do not change that. What changes is that what ran, and what came back, can be shown rather than asserted. If you have never added an MCP server to Claude Code before, this page assumes nothing: it starts at the settings, uses one hosted endpoint as the worked example, and ends with a test call you verify yourself.
Before you start: what you need
You need Claude Code running as you normally use it, the server address copied exactly, and one small test job. The address is https://zambo.dev/mcp. There is no API key to generate or store, and for a hosted remote server like this one there is nothing to install locally and no server process to keep running on your machine. That last point matters more than it sounds: a locally run MCP server is another dependency to install, update, and debug, while a hosted endpoint moves that maintenance to the provider. Your config holds one URL and stays one URL.
Pick the test job before you open any settings: something small, external, and checkable by eye. Look up a current price. Run a quick audit on a page you own. Fetch a fact you can compare against a source open in another tab. The test is how you will know the connection works, because a saved config entry is only a claim. A tool listing, a returned result, and a verified receipt are evidence.
Step by step: add the Zambo MCP server to Claude Code
Step 1. Open the MCP or connectors settings in Claude Code, or the MCP configuration it reads.
Step 2. Add a remote server with the URL https://zambo.dev/mcp. Name it zambo, so later sessions and teammates can tell at a glance what the entry is.
Step 3. Save the configuration. No API key is needed at any step.
Step 4. Reload the session, so it starts with the updated server list.
Step 5. Confirm the Zambo tools are listed in the session before you point the setup at real work.
Step 6. Run the small test call from the next section, and check its receipt.
Six steps, and only one value to get right: the URL. Keep it exactly https://zambo.dev/mcp for the default surface. The full surface variant, with ?full=true added, is covered below, and it is a deliberate choice for specific sessions, not the setting to leave on by accident. If your team shares a Claude Code setup, this same entry belongs in the shared configuration the same way, so every developer session on the project gets the same tools, the same memory store on the same account where you intend that, and the same receipt trail.
Your first tool call, and how to check it
In the reloaded session, ask for the test job in plain words and name the source: ask Claude Code to use Zambo to check the thing you prepared. The result should come back from the tool, with current data you can compare, and it should carry a receipt link. Open the receipt, then paste its ID into zambo.dev/verify and run the free check, no signup needed. A match confirms the saved result was not changed after the fact.
That thirty second loop is the whole verification story for this setup, and it is worth doing before any real job runs through the connection. If you want to preview the loop with zero setup, the live demo runs one real Zambo call and hands you its receipt in about 10 seconds, and a recorded day of work can be browsed at zambo.dev/today. The endpoint itself is at zambo.dev/mcp: one URL, no key, nothing local.
What you can do once Claude Code is connected
Tools. The connected client can call Zambo tools directly in the session: web and data checks, current prices instead of training data guesses, site audits, document and file work, lead finding with the leadsignal tool, and the everyday jobs that used to mean opening five tabs and copying results back into the conversation by hand. Ask for the job in plain words and let the client pick the tool, then read the result the way you would read any work product, critically and in full.
Memory. Zambo memory holds only facts you explicitly save, through memory_remember, and brings them back through memory_recall. Save a preference, a project rule, a price, or the current state of a long job once, and any AI connected to the same Zambo account can recall it later. The store does not import your conversations and it does not guess: if a fact is in there, someone put it there on purpose, and you can review and correct it the same way.
Handover. When a job is finished, package it as a handover: one link carrying the goal, the steps in order, the outputs, and any open items. The person receiving it opens the link with no account and clicks Accept or Flag. You cannot accept your own handover, which is exactly what makes the acceptance recorded on it worth showing later.
Receipts. Every tool call returns a verifiable receipt for its saved result. Anyone can check a receipt free at zambo.dev/verify with no signup, and the check confirms the saved result was not changed after the fact. It does not prove the result was right, and this page will not pretend otherwise. It proves what ran and what came back, unaltered, which is the half of accountability that a confident summary cannot supply on its own.
Default surface or full surface in a coding session
The default surface loads 35 tools: the everyday set for web and data work, checks and audits, file jobs, lead finding, memory, continuity, and handover. In a coding session that discipline pays for itself. Claude Code is already choosing among its own file, edit, and command tools; adding a focused external set keeps its picks fast and its reasoning about tools short. Most repository work, even work that reaches outside the repo several times a day, lives comfortably on the default surface for months.
The full surface is https://zambo.dev/mcp?full=true, all 153 tools in the session. Reach for it deliberately: an exploration day when you are mapping what the catalogue can do, a broad audit where the needed tool is not known in advance, or an agent build that routes across the whole set by design. Then return the configuration to the default URL. The failure mode to avoid is leaving full on everywhere and wondering why tool picks got slower and stranger: that is a toolbox problem, not a model problem, and the fix is the smaller surface.
Using the connection well on real repository work
Three uses tend to survive the first week and become habit. First, external facts at the moment of need: a library behaviour, a current price in a billing change, the live state of a page the code deploys. A tool call in the session beats a tab hunt and a paste, and it leaves a receipt in the record of the day. Second, session memory with intent: when the agent discovers a dead end, a naming rule, or a deploy gotcha, have it save that fact with memory_remember in one sentence. The next session, and the planning chat in Claude or ChatGPT on the same account, can recall it with memory_recall instead of paying the discovery cost twice.
Third, delivery with a record. When Claude Code work is for a client or a teammate, package the finished job as a handover: goal, steps, outputs, receipts, open items, in one link they open with no account and accept or flag. The acceptance matters because you cannot accept your own handover, so the decision on the record is genuinely theirs. Paired with your normal review and tests, which remain your job, the handover turns an agent session from a claim of done into a packet someone else can check line by line.
Troubleshooting: when the server does not show up
- The tools are not listed after reloading. Check the MCP configuration Claude Code reads: the entry must be a remote server with the URL exactly https://zambo.dev/mcp, saved in the configuration the session actually loads.
- The entry saved but calls fail. Reload the session once more and run the small test call again. A session started before the configuration changed can hold the old server list.
- Tool picks are slow or odd. Check whether the URL carries ?full=true. For daily coding, edit it back to the plain default URL and keep full for named jobs only.
- A call hits a usage limit. Receipt checking stays free at https://zambo.dev/verify/. A Day Pass covers 24 hours of full access for a heavy build day; see the cost section.
- You want the server removed. Delete the entry from the same MCP configuration. Nothing was installed locally, so removal is complete, and re-adding is the same one entry.
What it costs
Connecting the server costs nothing to set up, and checking a receipt is always free. When the connection becomes part of a real working day, a Day Pass at $0.99 covers 24 hours of full access: enough to connect in the morning, run real jobs through the day, save the facts that matter, and hand the finished work over with its record attached. Ongoing use is the Zambo Pass at $29 a month or $290 a year, and $19 a month for $ZAMBO holders. Humans pay by PayPal or card through PayPal; agents pay USDC on Base through x402. Read that pricing as the price of solving the connection problem properly, with tools, memory, and proof in one place, rather than as a charge for the paste itself, which takes a minute and costs nothing.
Honest limits
Three limits, stated plainly. First, a receipt proves a saved result was not changed after the fact; it never proves the code Claude Code wrote was correct, so reading diffs and running tests remains the quality gate, exactly as before. Second, every added server costs context and attention, Zambo included: if a project needs no outside tools, run it without the connection and add it when a named job appears. Third, this page describes the hosted Zambo endpoint from the inside; judge any other server you add by the same standard, one small call, verified by eye, before it touches real work. AER-1, the receipt format underneath, is an IETF Internet-Draft (draft-zambo-aer1-24), not a finalized standard.
Frequently asked questions
How do I add an MCP server to Claude Code?
Open the MCP or connectors settings in Claude Code, or the MCP configuration it reads, and add a remote server with the URL https://zambo.dev/mcp. No API key is needed. Reload the session, confirm the Zambo tools are listed, and run one small call you can verify by eye before real work. For the full catalogue in a session, use https://zambo.dev/mcp?full=true, which exposes all 153 tools instead of the 35 tool default surface.
Do I need to install anything locally to add an MCP server to Claude Code?
No local install is needed for Zambo, because the server is hosted. The entry is a remote server URL, https://zambo.dev/mcp, in the MCP configuration Claude Code reads: there is no server process to run on your machine, no package to keep updated, and no API key to store. When Zambo adds a tool, your next session has it. Locally run servers are a different shape and do need a command and a process.
What can Claude Code do once an MCP server is connected?
Alongside its normal coding work, Claude Code can call Zambo tools in the session: live web and data checks, current prices and documentation facts, site audits, lead work, and file jobs, with a verifiable receipt for each call. It can save project facts and work state with memory_remember and recall them with memory_recall in later sessions and in other connected AIs, and it can package finished work as a handover link a client can accept or flag.
Should I run the default or the full Zambo surface in Claude Code?
Run the default surface, https://zambo.dev/mcp with 35 tools, for focused daily coding: the model picks faster and the tool list stays readable in a busy session. Switch to the full surface, https://zambo.dev/mcp?full=true with all 153 tools, for exploration days and broad audits where you do not know in advance which specialist tool the job needs. Treat full as a workshop you visit for a named job, not the bench you code at every morning.
How much does it cost to add an MCP server to Claude Code this way?
Adding the server costs nothing to set up, and checking a receipt is free at https://zambo.dev/verify/ with no signup. For $0.99, a Day Pass covers 24 hours of full access, enough for a full day of recorded agent work. Ongoing use is the Zambo Pass at $29 a month or $290 a year, and $19 a month for $ZAMBO holders. The endpoint runs 153 tools in total with 35 on the default surface. Humans pay by PayPal or card through PayPal; agents pay USDC on Base through x402.