Skip to content
MCP server

Implicit Core

Local MCP server for addressing synthetic agent-environment state and checking provenance

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted
Runs
Locally
Version
1.0.0
Listing updated
1 day ago

About Implicit Core

Implicit Core is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.

Implicit Core is infrastructure for agent training and evaluation environments, where a large possible environment is addressed by coordinate and only the state an episode needs is loaded. The MCP server is a bounded local demo of that idea, running over stdio with a synthetic in-memory warehouse; no cloud or authentication is required.

It exposes eight tools for creating a synthetic universe of up to a million items, addressing a coordinate, materializing inventory or policy resources on demand, inspecting state and reading provenance hashes. Limits are strict: 16 universes and 64 addresses per process, 65,536 bytes per message, and state is discarded on restart. It has no sockets, shell or arbitrary file access.

Real environments are connected through the project's Python adapter layer rather than through MCP. The project is Apache-2.0 and installs with pip as implicit-ai; the README reports a 93.88% aggregate reduction in serialized state on its own earlier benchmark population and notes this is not RAM.

Install

Run in a terminal

claude mcp add implicit-core -- npx -y https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb

Add to claude_desktop_config.json

{
  "mcpServers": {
    "implicit-core": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "implicit-core": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "implicit-core": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "implicit-core": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "implicit-core": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb"
      ]
    }
  }
}

Tools this server provides (8)

Taken from the project documentation; the running server may differ.

implicit_create_universe
Create a synthetic in-memory universe of 1 to 1,000,000 items.
implicit_inspect_universe
Read known metadata for a universe.
implicit_address_experience
Create an address for a coordinate without loading pages.
implicit_materialize
Load the inventory or policy resource for a coordinate.
implicit_inspect_state
Read materialization metadata.
implicit_get_provenance
Read address, source version and hashes.
implicit_benchmark
Run the fixed public toy benchmark.
implicit_validate_adapter
Validate the built-in warehouse contract.

What it does

Synthetic universes
Create an in-memory environment of 1 to 1,000,000 items.
Selective materialization
Load inventory or policy for one coordinate only when requested.
Provenance
Read address, source version and hashes for what was loaded.
Built-in benchmark
Run a fixed public toy comparing eager and selective state.
Strict bounds
Capped universes, addresses and message size; no sockets or shell.
Local stdio
Runs offline via the implicit-mcp command.

Good for

  • Try the experience-layer idea before writing an adapter
  • Demonstrate on-demand state loading to a team
  • Check provenance output from a coding agent
  • Validate the built-in warehouse adapter contract

Pricing

Open source

Apache-2.0; free to install with pip

Prices change. Check zeitcow.github.io before you buy.

Implicit Core FAQ

How is it installed?

Run pip install implicit-ai in a fresh Python environment, then start implicit-mcp, which speaks JSON-RPC over stdio.

Does it need credentials?

No cloud or authentication is required.

Can it connect to my own environment?

Not through MCP; real environments use the Python adapter layer.

What licence is it?

The README states Apache-2.0.

Is Implicit Core free?

Yes. Implicit Core is open source, so you can use and self-host it for free.

How do I add Implicit Core to Claude Code?

Run this command in your terminal: claude mcp add implicit-core -- npx -y https://github.com/zeitcow/implicit-core/releases/download/v1.0.0/implicit-core-mcp-1.0.0.mcpb

What are the best alternatives to Implicit Core?

The closest alternatives to Implicit Core are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

Alternatives to Implicit Core

  1. REA

    MCP server

    CLI and MCP server that lets coding agents reverse engineer apps and binaries

    Runs locally
    • Open source
    • 464K GitHub stars
    • Developer tools
  2. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  3. GitHub

    MCP server

    Official MCP server connecting AI assistants to GitHub repos, issues and workflows

    Not checked yet
    • Remote
    • Open source
    • 34.5K GitHub stars
    • Developer tools
  4. MCP toolkit giving coding agents symbol-level code retrieval and editing

    Runs locally
    • Open source
    • 31.5K GitHub stars
    • Developer tools
  5. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  6. Ouroboros

    MCP server

    Spec-first layer that pins requirements and verifies results of AI coding agent work

    Runs locally
    • Open source
    • 6.4K GitHub stars
    • Developer tools