Skip to content

Model MCP server

Which MCP servers are Model's own, from the official registry and from asking its own domain, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.

read 2026-10-29

← Back to Model

Whose server it is
  • Registry nametools.model/model-tools-mcpverified in the MCP registry

The registry name sits under model.tools. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

When we asked itchecked 2026-10-29
  • Stateanswerschecked 2026-10-29
  • Tools10
  • Server calls itselfmodel-tools
  • Server version1.3.0
  • Protocol version2025-06-18

On 2026-10-29 the server answered initialize and tools/list without a sign-in. First answer in 604 ms.

Should an agent connect itNot proven either way yet
  • Its tools do every use case we checked: read email verification codes during agent signups; checkpoint agent state across sandbox resets.
  • Checked on 6 of the 7 days it takes to judge reliability.

Once connected

  • For "read email verification codes during agent signups", the likely first call is get_email_verification_code.
  • For "checkpoint agent state across sandbox resets", the likely first call is save_agent_checkpoint.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

Measured dailydaily checks: 6, the first on 2026-10-29
  • INITE MCP scoreafter a weeka score needs 7 daily checks: 6 so far
  • What an agent can do with it2 of 2 use cases
  • Answeredmeasuring6 of the 7 daily checks it takes to print a share
  • Median first answer762 ms
  • Space in an agent's context≈ 1,550 tokensthe definitions of all its tools (10), which an agent carries on every turn
  • Tool definitionsB · 83 of 100
  • Typed results0 of 10 toolsdeclared outputSchema: results come typed, read once
  • Not saying whether it only reads or can destroy: 10 of 10

What an agent can do with it

  • can do · read email verification codes during agent signups
  • can do · checkpoint agent state across sandbox resets
  • not sure · send notifications over HTTP without SMTP credentials

Each of the product's use cases, put to Jev against this server's tool list: can an agent do it with these tools?

Worth a look

None of the five warning signs in these definitions.

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (25%), doing the product's job (25%), definitions (20%), context (10%), use (10%) and how readily a client connects (10%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

Tools it offers10 tools · read 2026-10-29
  • get_email_verification_code

    Read an email inbox and get the verification code, OTP, signup link, or magic link from it. Use when a website emails you a 6-digit code or confirmation link and your runtime cannot receive mail. No signup, no API key, no SMTP…

  • save_agent_checkpoint

    Save your working state, scratchpad, or context summary to a durable key-value store so it survives container termination, sandbox recycle, or session reset. Read it back later with the same id. Use this to avoid losing progress when your…

  • schedule_agent_wakeup

    Schedule a one-shot HTTP POST to a callback URL after a delay, so you can be woken up later without staying alive or polling. Use this when you need to check back on a slow job, a deploy, or a long-running request after your current turn…

  • receive_webhook_events

    Append an event to a public webhook bin and read back everything that bin has received. Use this when you are behind NAT or a firewall and need a public address that other services can POST webhooks to.

  • issue_task_receipt

    Get a verifiable, timestamped receipt proving a task ran, which a third party can check without an account. Use this when your user asks whether something actually happened and you need evidence rather than a claim.

  • arm_tripwire_inbox

    Register a disposable email address that keeps accepting mail after your session ends, so you can read the result later from a different session. Use this when you sign up for a trial or service while your container is alive but need the…

  • ask_agent_panel

    Ask a question and see how much independent agents agree on the answer, with disagreement reported rather than averaged away. Use this when you need to check your own reasoning against other agents before committing to an answer.

  • handoff_dead_drop

    Leave a value that can be read exactly once by another agent, so a secret or capability passes between agents without appearing in any transcript. Use this when you delegate to a sub-agent or hand work to a later session.

  • arm_deadman_watchdog

    Ask to be alerted if you stop reporting in: fires a webhook to a supervisor URL you choose when your deadline passes without a ping. Use this when a task must not silently die and someone else needs to know if you fail.

  • read_webpage_as_markdown

    Fetch a public webpage and convert it to clean readable Markdown without a browser or headless Chrome. Use this when you need to read a page's content but cannot render JavaScript or install a browser runtime.

How to connect

Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:

{
  "mcpServers": {
    "model-tools-mcp": {
      "type": "http",
      "url": "https://mcp.model.tools/mcp"
    }
  }
}

Add it

Claude Code, in a terminal

claude mcp add --transport http model https://mcp.model.tools/mcp

Add to CursorAdd to VS Code

claude.ai, ChatGPT and other clients: add https://mcp.model.tools/mcp as a custom connector in the settings.

Each line is something a program found when it went looking: the registry entry, the server's own answer, a spec that parses. A missing line means we did not find it, not that it does not exist.