ninetrix

Build and deploy AI agents as Docker containers. Define your agent in YAML, ship it anywhere Docker runs.

pip install ninetrix

Quickstart

# Scaffold a new agent
ninetrix init --name my-agent --provider anthropic

# Build the container image
ninetrix build --file ninetrix.yaml

# Run it interactively
ninetrix run --file ninetrix.yaml

Multi-agent crews

# Start all agents on a shared Docker network
ninetrix up --file ninetrix.yaml

# Trigger the orchestrator
ninetrix invoke --agent orchestrator -m "Research Python history and write a summary"

# Stream logs from all agents
ninetrix logs --file ninetrix.yaml

# Visualize the execution trace
ninetrix trace --thread-id <id>

# Tear down
ninetrix down --file ninetrix.yaml

ninetrix.yaml

agents:
  orchestrator:
    metadata:
      role: "Research Orchestrator"
      goal: "Coordinate search and synthesis"
    runtime:
      provider: anthropic
      model: claude-sonnet-4-6
    tools:
      - { name: search, source: mcp://duckduckgo }
    collaborators: [researcher, writer]
    governance:
      max_budget_per_run: 1.00
      human_approval: true
    triggers:
      - type: webhook
        endpoint: /run

  researcher:
    runtime: { model: claude-haiku-4-5-20251001 }
    tools:
      - { name: search, source: mcp://duckduckgo }
      - { name: files,  source: mcp://filesystem }

  writer:
    runtime: { model: claude-sonnet-4-6, temperature: 0.7 }
    tools:
      - { name: files, source: mcp://filesystem }

Commands

Environment variables

Requirements

• Python 3.10+
• Docker

License

MIT