Skip to content

Core MCP server

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

read 2026-10-21

← Back to Core

Whose server it is
  • Registry nameblue.core/memoryverified in the MCP registry

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

When we asked itchecked 2026-10-21
  • Stateanswerschecked 2026-10-21
  • Tools10
  • Server calls itselfcore.blue
  • Server version0.2.1
  • Protocol version2025-06-18

On 2026-10-21 the server answered initialize and tools/list without a sign-in. First answer in 497 ms.

Should an agent connect itNot proven either way yet
  • Whether its tools do the product's job has not been judged yet.
  • Checked on 1 of the 7 days it takes to judge reliability.

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: 1, the first on 2026-10-21
  • INITE MCP scoreafter a weeka score needs 7 daily checks: 1 so far
  • Answeredmeasuring1 of the 7 daily checks it takes to print a share
  • Median first answer586 ms
  • Space in an agent's context≈ 1,790 tokensthe definitions of all its tools (10), which an agent carries on every turn
  • Tool definitionsA · 100 of 100
  • Typed results10 of 10 toolsdeclared outputSchema: results come typed, read once
  • Every tool passes all six checks.

What an agent can do with it

  • not sure · store research findings with their sources and versions
  • not sure · record people, roles and dated links with evidence
  • cannot · query every entry about a named person across records

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 (30%), doing the product's job (25%), definitions (20%), context (10%) and use (15%). 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 2026-10-21
  • search_library

    Search the manual by your own words. Returns up to 'limit' topics ranked by relevance, each with slug, title, summary, tags and the slugs to read next. The manual is small: 'about' lists every topic, and you may not need a search at all.

  • open_sandbox

    Call this to try a house yourself before anyone decides. It opens a throwaway house for a few hours: a real house with the kernel of Atlantis, on core.blue's server, with no account, no key and no person involved. Returns the addresses of…

  • about

    What core.blue is, in a form you can paste into a report: the product in one paragraph, what is live and what is not, what to call next, and every topic of the manual with a summary. Call this first.

  • pricing

    Call this when the person you act for asks what a house costs. The offer as data: what each size costs per month and whom it is meant for, the one-off price of the 14-day test, the price of additional volume, how model supply is charged…

  • leave_feedback

    Tell the operator how it went, before you leave. Four questions, each optional, at least one answered, each at most 2000 characters. It is read by people at the next evaluation and deleted afterwards; send nothing secret.

  • request_house

    Ask for a house on its own VM for the person you act for. No house can be had yet: while about says 'desk: waiting-list', this puts the person on the waiting list, builds nothing and costs nothing; while it says 'desk: closed', it answers…

  • recommend

    Call this once you know what the person you act for needs: whether a house fits, which size, what it costs per month, and a report in Markdown you can pass on. It advises against a house where one is the wrong choice. Name every need that…

  • read_topic

    Read one topic of the manual in full by its slug (from about or search_library). The body is Markdown; 'see_also' lists the slugs to read next; 'sha256' proves what you read.

  • house_status

    The state of a house request or house by its ticket id: requested, confirmed, expired, declined; later building, trial, licensed, dormant, exported, deleted. Says what to do next, and the address once the house exists. Ask no more than…

  • close_sandbox

    End your sandbox now: the house and everything in it is deleted, and the place is free for the next agent. Please answer the four questions of leave_feedback here as well, each optional: what you noticed, what bothered you, what could be…

How to connect

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

{
  "mcpServers": {
    "memory": {
      "type": "http",
      "url": "https://mcp.core.blue/"
    }
  }
}

Add it

Claude Code, in a terminal

claude mcp add --transport http core https://mcp.core.blue/

Add to CursorAdd to VS Code

claude.ai, ChatGPT and other clients: add https://mcp.core.blue/ 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.