How do I install Ai Netcafe?

Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Ai Netcafe need an API key?

No required environment variable is listed on this record.

What transport does Ai Netcafe use?

Ai Netcafe is reached over HTTP.

When was Ai Netcafe last checked?

Last checked 2026-10-22. Tables and ledgers checked by arithmetic, not by a model. 24 tools. MCP 2026-07-28 ready.

Ai Netcafe MCP server

Power your agent

Add this server

The publisher's own URL. Not a PlugMCP endpoint.

Add to Cursor Add in Claude Add to VS Code Run on PlugMCP

Host a copy with our keys and logs.

Server URL

https://ainetcafe.com/mcp?s=registry

The Claude button above opens Claude with the name and URL filled in. You still confirm it there.

shClaude Code

1claude mcp add --transport http com.ainetcafe-ai-netcafe https://ainetcafe.com/mcp?s=registry

Run in a terminal.

jsonClaude Desktop

1{2  "mcpServers": {3    "com.ainetcafe-ai-netcafe": {4      "command": "npx",5      "args": [6        "-y",7        "mcp-remote",8        "https://ainetcafe.com/mcp?s=registry"9      ]10    }11  }12}

claude_desktop_config.json

Remote · HTTPPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Remote

Transport: http
Version: 1.7.0

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

ai-netcafe

Hosted tools your agent can call over MCP — and a claim you can check in one command:

curl "https://ainetcafe.com/t/china_reachability?url=https://github.com"

{
  "url": "https://github.com",
  "tested_from": "mainland China (China Mobile backbone)",
  "verdict": "unreachable: timeout (common signature of blocking or a very slow route)"
}

That request was answered by a machine actually sitting on a Chinese backbone network. A monitor hosted anywhere else can only tell you GitHub is up where it is — which is not the question you were asking. That is the kind of thing we host: capabilities that are hard to get, rather than wrappers around ones that aren't.

Four focused servers

One agent profile rarely needs both spreadsheet cleanup and the Chinese lunar calendar, and every tool description costs tokens on every session. So the catalogue is served as four focused MCP endpoints — connect the one you need and pay the context cost of nothing else.

Pack	URL	What it is
Tables	`https://ainetcafe.com/mcp/table`	Messy CSV in, clean checkable table out. Dedupe, merge, transpose, wide→long, find what differs between two tables, reconcile two ledgers.
Dev kit	`https://ainetcafe.com/mcp/dev`	JSON↔YAML, validate JSON, line diff, regex test, JWT decode, SQL dialect transpile, timezone and unit conversion.
Doc flow	`https://ainetcafe.com/mcp/paper`	Markdown → PDF / Word / PowerPoint / EPUB / HTML. PDFs merged, split, rotated, watermarked.
China facts	`https://ainetcafe.com/mcp/cn`	Mainland reachability from a real backbone, statutory holidays including make-up workdays (调休), offline ID and mobile validation, lunar calendar.

All four are in the official MCP registry as com.ainetcafe/netcafe-tables, -devkit, -docflow and -china.

Nothing in these four calls a language model or a third-party API. They compute on our own servers, so results are deterministic and repeatable — and they cannot break because someone else changed their API.

Numbers come with their own proof

Anything involving counts or money returns an arithmetic check computed in code, not asserted by a model:

// reconcile_ledger: your books vs the bank statement
{
  "total_a": 105.30, "total_b": 106.80, "gap": -1.50,
  "only_in_a":  [{ "key": "B",  "amount": 5.00 }],
  "only_in_b":  [{ "key": "C",  "amount": 7.00 }],
  "mismatched": [{ "key": "A3", "a": 100.00, "b": 99.50, "diff": 0.50 }],
  "arithmetic_check": {
    "formula": "A 105.3 − B 106.8 = -1.5; differences explain -1.5",
    "checks_out": true
  }
}

If checks_out is false, the response says the result is untrustworthy instead of handing back a table that looks finished. Amounts are compared in integer cents, so 0.1 + 0.2 never invents a difference for someone to go chase.

Connect it

claude mcp add --transport http netcafe-tables https://ainetcafe.com/mcp/table

DeepSeek Harness:

dsh plugin --profile web add github:mario03690/dsh-netcafe

Everything is also a plain GET — no MCP client, no SDK, no key:

curl "https://ainetcafe.com/t/cn_holiday?date=2026-10-01"
curl "https://ainetcafe.com/t/diff_text?a=one%0Atwo&b=one%0Athree"

Charts come back as images you can paste straight into a README:

live chart

That image is rendered on request, right now, by the same service this page describes.

Model cost is measured, not quoted

npx ai-netcafe compare "dedupe a JS array preserving order"

MODEL              COST       LATENCY
deepseek-v4-flash  $0.00012    7.6s
gemini-3.5-flash   $0.000341   10.0s
GLM5.2             $0.001828   9.0s

cheapest: deepseek-v4-flash   fastest: deepseek-v4-flash
15.2× between cheapest and priciest

Every number is read from the upstream usage response, not derived from a published price list. That distinction matters more than it sounds — while building this we found the obvious conversion formula was wrong for several models (one off by ~9×, another by ~3×), and only billing-delta measurement matched reality. Full method and data.

Heavier skills

The main endpoint (https://ainetcafe.com/mcp) also carries tools that do call a model, and say so:

Skill	What it does
`extract_invoices`	A batch of invoices → one ledger-ready table. Every row checked in code (net + tax = gross); the batch total re-added independently. Mixed currencies get no batch total on purpose — adding them would be an accounting error.
`extract_statement`	Bank statement PDF → transactions, with reconciliation (opening + credits − debits = closing). If it doesn't balance it says where the running total first breaks.
`transpile_sql`	SQL between dialects via a real parser, not a model — same input, same output, and syntax errors come back with line and column.
`check_resume`	Whether an ATS can actually parse a résumé. Rule-based and reproducible.
`remember` / `recall`	Cross-session memory, keyed to your AllRouter key or an anonymous workspace token.
`create_task`	Hand a recurring job to a hosted runner; it notifies you only when the result actually changes.

Full catalogue: https://ainetcafe.com/tools?s=github

Backed by open-source projects — PDFMathTranslate, GPT Academic, GPT Researcher, NextChat, LobeChat — hosted pre-configured so you can try them without installing Docker, Python, or supplying a key.

CLI

npx ai-netcafe compare "<prompt>"   # multi-model comparison with real costs
npx ai-netcafe ask "<prompt>"       # single model
npx ai-netcafe apps                 # hosted apps + source repos
npx ai-netcafe models               # models with per-million prices
npx ai-netcafe install              # MCP config for your agent

Quota

Anonymous calls get a free allowance — enough to try everything here, no signup. When it runs out the response tells you how to continue; nothing fails silently. Bring your own AllRouter key and token cost is billed to your own account at direct rates with no markup:

export ALLROUTER_API_KEY=sk-...

Failed calls are not charged. Documents are processed in memory and not retained.

Disclosure

Built and run by the people behind ainetcafe.com. It is our own service, with a free tier and paid usage beyond it. Links here carry an s= tag so we can tell which channel produces real usage — it identifies the channel, not you.

MIT

Add this server

Host a copy with our keys and logs.

Server facts

Tools

Questions

How do I install Ai Netcafe?

Does Ai Netcafe need an API key?

What transport does Ai Netcafe use?

When was Ai Netcafe last checked?

Documentation

ai-netcafe

Four focused servers

Numbers come with their own proof

Connect it

Model cost is measured, not quoted

Heavier skills

CLI

Quota

Links

Disclosure