Multi-Agent Workspaces

cortex-engine supports multiple agents in a single workspace, each with isolated memory namespaces.

Adding Agents

npx fozikio init my-workspace
cd my-workspace
npx fozikio agent add researcher
npx fozikio agent add writer

Each agent gets its own memory namespace. The researcher agent cannot see writer memories, and vice versa.

Serving Specific Agents

# Serve all agents (default namespace)
npx fozikio serve

# Serve a specific agent
npx fozikio serve --agent researcher

Claude Code Integration

Add multiple agents to your Claude Code config:

{
  "mcpServers": {
    "researcher": {
      "command": "npx",
      "args": ["fozikio", "serve", "--agent", "researcher"]
    },
    "writer": {
      "command": "npx",
      "args": ["fozikio", "serve", "--agent", "writer"]
    }
  }
}

How Isolation Works

Each agent's memory is stored in a separate namespace within the same storage provider. This means:

• SQLite: Separate tables per agent within the same database file
• Firestore: Separate document collections per agent

Agents share the same embedding engine but have completely independent memory stores.