Skip to content
MCP server

MCP Server

Typed record store with hybrid search and RAG for AI agents, scoped per customer

Runs locallyStatus and history
Install
Pricing
Paid
Platforms
CLI, API
Runs
Locally
Version
0.19.0
Listing updated
3 days ago

About MCP Server

MCP Server is a paid MCP server for databases. It runs locally on your machine and is started by your AI client.

Memory for an agent usually means fuzzy recall; this server pairs exact lookups with semantic search over the same isolated, per-customer index. Records follow schemas you declare (fields, validation, references, sensitivity), documents are ingested and indexed automatically, and one search call can run in keyword, semantic or hybrid mode over both.

The npm package runs locally over stdio and talks to the Vectros backend-as-a-service, so a key is required. The registry entry recommends a least-privilege scoped key (ssk_...), which the Vectros CLI bootstrap command can mint and merge into Claude Desktop, Cursor, Cline or Claude Code config. A tool allow-list variable restricts which of the 26 tools register. The server itself makes no web calls, though stored agents it runs can search or fetch the web when their definition lists those tools.

The platform behind it is credit-based: one credit is a cent, reads draw on a free allowance, and model inference is billed from a separate prepaid balance. The package is Apache-2.0 licensed.

Install

Run in a terminal

claude mcp add mcp-server --env VECTROS_API_KEY='<VECTROS_API_KEY>' -- npx -y @vectros-ai/mcp-server

Add to claude_desktop_config.json

{
  "mcpServers": {
    "mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@vectros-ai/mcp-server"
      ],
      "env": {
        "VECTROS_API_KEY": "<VECTROS_API_KEY>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@vectros-ai/mcp-server"
      ],
      "env": {
        "VECTROS_API_KEY": "<VECTROS_API_KEY>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@vectros-ai/mcp-server"
      ],
      "env": {
        "VECTROS_API_KEY": "<VECTROS_API_KEY>"
      }
    }
  }
}

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

{
  "mcpServers": {
    "mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@vectros-ai/mcp-server"
      ],
      "env": {
        "VECTROS_API_KEY": "<VECTROS_API_KEY>"
      }
    }
  }
}

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

{
  "mcpServers": {
    "mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@vectros-ai/mcp-server"
      ],
      "env": {
        "VECTROS_API_KEY": "<VECTROS_API_KEY>"
      }
    }
  }
}

Configuration

  • VECTROS_API_KEYVectros API key. Use a least-privilege scoped key (ssk_...); mint one with `npx -y @vectros-ai/cli bootstrap`. Required for a registry install, which cannot assume any other tooling is present. If you also install the vectros CLI (0.9.0+), the server can instead resolve the key from its keyring and this can be left unset.Required
  • VECTROS_KEYRING_ALIASOptional. When VECTROS_API_KEY is unset and the vectros CLI is installed, resolve this keyring entry instead of the active one.
  • VECTROS_API_BASE_URLOverride the API base URL (default https://api.vectros.ai). Must be an official *.vectros.ai host.
  • VECTROS_MCP_TOOLSAllow-list: comma-separated names of the only tools to register (default: all 26 tools).

What it does

Typed records
Declared schemas with versioned writes and optional optimistic concurrency.
Hybrid search
One call over records and documents in keyword, semantic or hybrid mode.
Document ingestion
Inline or uploaded documents, organised in folders and indexed automatically.
Grounded answers
Question answering over indexed content with citations to sources.
Scoped keys
Least-privilege keys; every caller's data is isolated by default.
Tool allow-list
An environment variable limits which of the 26 tools are registered.

Good for

  • Give a coding or support agent durable memory with exact and semantic recall
  • Search a customer's records and uploaded documents together
  • Run stored agents from an MCP client
  • Keep each tenant's data separate in a multi-customer AI product
  • Answer questions grounded in ingested documents with citations

Pricing

Paid

Credit-based pay-as-you-go, 1 credit = $0.01; free read allowance; plan prices not published

Prices change. Check vectros.ai before you buy.

MCP Server FAQ

How do I get an API key?

Run npx -y @vectros-ai/cli bootstrap, which mints a scoped key and updates your client config; you can also paste an ssk_ key.

Which clients does it work with?

The README lists Claude Desktop, Cursor, Claude Code, Cline, Continue and VS Code.

Can I limit what the agent can do?

Yes. Use a scoped key and the VECTROS_MCP_TOOLS allow-list to register only chosen tools.

What does the backend cost?

Pay-as-you-go credits at one cent each, with a read allowance on every plan; inference uses a separate prepaid balance.

Is MCP Server free?

No. MCP Server is a paid product.

How do I add MCP Server to Claude Code?

Run this command in your terminal: claude mcp add mcp-server --env VECTROS_API_KEY='<VECTROS_API_KEY>' -- npx -y @vectros-ai/mcp-server

What are the best alternatives to MCP Server?

The closest alternatives to MCP Server are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and Midplane.

Alternatives to MCP Server

  1. 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
  2. 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
  3. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  4. Local-first long-term memory layer for AI agents, served over MCP

    Not checked yet
    • 5 tools
    • Remote
    • Open source
    • 1.3K GitHub stars
    • Knowledge and memory
  5. Midplane

    MCP server

    Postgres gateway for AI agents with table grants, column masking and approval for writes

    Runs locally
    • Open source
    • 5 GitHub stars
    • Databases
  6. note1

    MCP server

    MCP server giving AI assistants search and summaries across your note1 meetings

    • 4 tools
    • Remote
    • Free trial
    • 1 GitHub star
    • Productivity