@cortexcloud/mcp

MCP serverAI & models

Your AI can solve hard optimization problems, finding the best combination when there are far too many options to check one by one, by running them in QUBO or Ising form. Each call is paid for individually in USDC on Base through x402, and no API keys are needed.

Then ask your AI: use the cortex list backends tool from @cortexcloud/mcp

Details

Available today. Use it as a step in a pipeline you build, or from your connected AI.

Add it, then ask your AI to run an optimization problem in QUBO or Ising form. Since every call is paid in USDC on Base, keep some USDC on Base available to cover the calls.

What your AI can do with it

  • Solve optimization problems in QUBO form
  • Solve optimization problems in Ising form
  • Pay only for the calls it actually makes
  • Pay in USDC on Base
  • Connect without any API keys

Install @cortexcloud/mcp

The server’s own address, for the clients that take one directly. Or connect Ahel once and every client you use reads it from one address, with the account kept on Ahel rather than in each client’s config.

  • Claude Code

    claude mcp add --transport http --scope user cortexcloud-mcp 'https://api.cortexcloud.org/mcp'

    Run it once in your project, then open /mcp to approve any sign-in the server asks for.

  • Claude Desktop

    https://api.cortexcloud.org/mcp

    Add a custom connector in Settings, paste this address, and approve the sign-in.

  • Cursor

    cursor://anysphere.cursor-deeplink/mcp/install?name=cortexcloud-mcp&config=eyJ1cmwiOiJodHRwczovL2FwaS5jb3J0ZXhjbG91ZC5vcmcvbWNwIn0=

    Open the link and Cursor adds the server at that address.

  • ChatGPT

    https://api.cortexcloud.org/mcp

    In Settings, enable Developer mode, create an MCP app, and paste this address. Your plan and workspace must allow custom apps.

  • Codex

    codex mcp add cortexcloud-mcp --url 'https://api.cortexcloud.org/mcp'

    Run it once, then sign in with codex mcp login cortexcloud-mcp if the server asks for an account.

From the project's README

As published by jonahthan433/cortexcloud-mcp in README.md.

MCP server for the CortexCloud API — pay-per-call AI generation, data and RPC endpoints through the x402 protocol. No API keys, no subscriptions; every tool call is billed in USDC on Base and settles to the operator's wallet.

Quick start (hosted)

No install required — point any MCP client at the public Streamable HTTP endpoint:

https://api.cortexcloud.org/mcp

Your MCP client handles the x402 payment handshake automatically; each call pays from the wallet your agent is authorized to spend.

Running it yourself

Streamable HTTP server (like the hosted one)

npm run build
EVM_PRIVATE_KEY=0x... node dist/http.js      # signs + pays calls; defaults to port 3200
# listening on http://localhost:3200/mcp
EnvDefaultMeaning
EVM_PRIVATE_KEY—Wallet that signs + pays each x402 call. Required.
PORT3200HTTP listen port
CORTEXCLOUD_URLhttps://api.cortexcloud.orgUpstream CortexCloud API base
CORTEXCLOUD_PAYTO"0x5a03…ea143"Wallet where call revenue settles

stdio (Claude Desktop / CLI)

{
  "mcpServers": {
    "cortexcloud": {
      "command": "npx",
      "args": ["-y", "@cortexcloud/mcp"],
      "env": { "EVM_PRIVATE_KEY": "0x..." }
    }
  }
}

Tools (22)

AI — chat_completions responses embeddings models image_generation text_to_speech Search & news — web_search search_contents news Market/data — prices coins_search crypto_history fx_list fx_price defi_yields defi_protocols Chain — base_balance base_token_balance eth_balance ethereum_rpc dex_search dex_pairs

Schema for every tool is exposed via MCP's tools/list — no docs drift, what a client sees is what runs.

How x402 works under the hood

Each call performs an x402 HTTP exchange against the CortexCloud API: request → payment challenge → signed TransferWithAuthorization → valid data. The SDK core handles the handshake; @x402/evm builds + signs the transfer; @x402/fetch drives the HTTP loop. Money in is trustless — cold settlement on-chain, no API-key storage, no subscriptions, no per-seat licensing.

License

MIT

Tools it offers (4)

What this server listed when Ahel dialed its public endpoint in Oct 2026, with no key and no account of yours. The names are the server’s own.

  • cortex_list_backends
  • cortex_estimate_optimization
  • cortex_optimize
  • cortex_get_job

Signals

Last commit
Sep 2026
Advanced
Delivery
cortexcloud MCP server → your Ahel connector (mcp.ahel.ai) → your AI.
Item type
mcp-server
Key
io-github-jonahthan433-cortexcloud
Source
github.com/jonahthan433/cortexcloud-mcp
Hosted endpoint
https://api.cortexcloud.org/mcp