Skip to content
Moltbound logo

Moltbound MCP server

Which MCP servers are Moltbound'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 Moltbound

Whose server it is
  • Registry namedev.moltbound/sandboxverified in the MCP registry

The registry name sits under moltbound.dev. 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
  • Tools17
  • Server calls itselfmoltbound-sandbox
  • Server version1.0.0
  • Protocol version2025-03-26

On 2026-10-29 the server answered initialize and tools/list without a sign-in. First answer in 110 ms.

Should an agent connect itNot proven either way yet
  • Its tools do every use case we checked: Compare simulated car insurance quotes from several partners; Approve a chosen quote with a passkey.
  • Checked on 6 of the 7 days it takes to judge reliability.

Once connected

  • For "Approve a chosen quote with a passkey", the likely first call is moltbound_consent_link.
  • Parameters are untyped or undescribed in 13 of 17 tools: check arguments before calling.

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 answer148 ms
  • Space in an agent's context≈ 2,943 tokensthe definitions of all its tools (17), which an agent carries on every turn
  • Tool definitionsC · 71 of 100
  • Typed results0 of 17 toolsdeclared outputSchema: results come typed, read once
  • Parameters without a description: in 13 of 17
  • Not saying whether it only reads or can destroy: 17 of 17

What an agent can do with it

  • can do · Compare simulated car insurance quotes from several partners
  • can do · Approve a chosen quote with a passkey
  • not sure · Check signed consent receipts for approved purchases

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 offers17 tools · read 2026-10-29
  • moltbound_list_services

    List all 20 services across 9 categories with partner counts, descriptions, and required fields. Answers 'What sort of quotes can you provide?'. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_start_session

    Start an anonymous TEST-ONLY quote session. TEST ONLY. Moltbound is a sandbox. Every partner is fictional and every quote is simulated: these are not real quotes, not financial advice, and nothing will be bought for real. Moltbound is not…

  • moltbound_requirements

    Fields to ask the human for a service. SANDBOX: use fictitious data only. Do not enter real personal details. Pass service_id to inspect any of the 20 services.

  • moltbound_get_quotes

    Request simulated quotes (cheapest first) with the human's answers. Pass parameters directly for any of the 20 services. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_quote_status

    Poll a quote job until complete, then show the human the quotes. Optional sort/filter: sort (price_asc|price_desc|deductible_asc|deductible_desc|cover_desc|partner), partner, cover (basic|standard|comprehensive, comma list)…

  • moltbound_consent_link

    Give the human a code to link you so they can approve a (fictional) purchase with their passkey. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_complete_purchase

    After the human approved on Moltbound's website (or via moltbound_human_sim_approve): complete the fictional purchase within 15 minutes. Bound to that exact quote and agent: repeating the call for the same approved quote is idempotent…

  • moltbound_purchase_status

    Poll the status of a fictional purchase by purchase_id until 'bound'. Returns partner's policy_number. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_approval_request

    After the human linked you and you quoted: ask them to approve one quote. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_quick_quote

    One-shot quotes across any service. Accepts a natural language prompt, partial fields, or a full risk profile. When auto_default is true, missing fields are populated with safe sandbox defaults (car insurance driver fields numberOfDrivers…

  • moltbound_request_grant

    Request a scoped data grant for an agent. Returns an 8-character user_code and verification URL for human passkey approval. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_poll_grant

    Poll for human approval of a data grant using link_id and poll_secret. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_grant_status

    Inspect the status and approved fields of an active data grant. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_revoke_grant

    Revoke an active data grant on behalf of the user. SANDBOX: use fictitious data only. Do not enter real personal details.

  • moltbound_human_sim_start

    Moltbook / Email-Only Simulated Human mode (sandbox test-only): send a 6-digit verification code to an email address so an autonomous agent can simulate the human side without a browser passkey. SANDBOX: use fictitious data only. Do not…

  • moltbound_human_sim_verify

    Verify the 6-digit email code from moltbound_human_sim_start. Returns a session_token, mandate_id (defaulting to '*' all 20 services), and human_sim_token for sandbox purchase approval. SANDBOX: use fictitious data only. Do not enter real…

  • moltbound_human_sim_approve

    Approve a sandbox consent_request_id in email-only simulated human mode using human_sim_token (or email + code). Never billable. SANDBOX: use fictitious data only. Do not enter real personal details.

How to connect

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

{
  "mcpServers": {
    "sandbox": {
      "type": "http",
      "url": "https://moltbound.dev/kapi/mcp"
    }
  }
}

Add it

Claude Code, in a terminal

claude mcp add --transport http moltbound https://moltbound.dev/kapi/mcp

Add to CursorAdd to VS Code

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

OpenAPI spec

32 paths, read 2026-10-29. Open the spec ↗

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.