Datapacks

The isolation unit for agent data in Meko

A datapack is Meko's fundamental unit of isolation. Each datapack provides a self-contained environment for an agent, bundling together a dedicated database, MCP server endpoint, and access credentials.

What's in a datapack

When you create a datapack, Meko provisions:

  • An MCP server endpoint. A unique URL that your agent uses to discover and interact with the datapack.
  • An MCP server API key. For authenticating MCP requests.
  • A dedicated database. An isolated YugabyteDB (YSQL) database with pgvector enabled, scoped to your agent.
  • Database credentials. A username, password, and connection string for direct PostgreSQL access.

Create a datapack

Using the CLI:

meko datapack create --name my-datapack

This returns the datapack's MCP server URL, API key, and database connection details:

{
  "datapack_id": "e02a7107-c79e-451d-b360-9cd1e13f175d",
  "agent_id": "18311fce-36c5-40fe-bd45-67f5913b0b52",
  "name": "my-datapack",
  "mcp_server": {
    "url": "https://mekodev.com/<account_id>/mcp",
    "api_key": "318ce1c589d08edb11e24b0a781fe675"
  },
  "db_access": {
    "user_name": "db_my-datapack",
    "password": "d28a0d1498c1ab9692184dd9c9cd1a79",
    "conn_str": "postgresql://db_my-datapack:d28a0d14...@host:5433/account_id_my-datapack"
  }
}

Isolation model

Each datapack is fully isolated:

  • Data isolation. Separate database per datapack. One agent cannot access another's data.
  • Credential isolation. Unique credentials per datapack.
  • MCP isolation. Each datapack's MCP server only exposes that datapack's tools and data.

Connect via MCP

Each datapack exposes an MCP server that agents use to interact with their data:

{
  "mcpServers": {
    "my-datapack": {
      "command": "meko",
      "args": ["mcp", "serve", "--name", "my-datapack"]
    }
  }
}

Add this configuration to your coding agent or agent framework to connect.

For the hosted MCP URL and API key (from the portal or meko datapack output), register multiple clients at once with Connect your AI client.

Manage datapacks

Command Description
meko datapack list List all your datapacks
meko datapack status --name <n> View datapack summary
meko datapack delete --name <n> Permanently delete a datapack

See the CLI reference for full details.