KDeps

The AI Appliance Builder.

AI agents as code. Declarative LLM orchestration for specialized and coordinated multi-agent systems. Compose workflows across models, APIs, and data with fully defined control flow.

Get Started

Why kdeps?

Registry

workflow.yaml

run:
  chat:
    model: llama3.2:1b
    prompt: "Summarize: {{ get('q') }}"

  apiResponse:
    success: true
    response:
      data: "{{ get('chat') }}"

# No glue code. No legacy code.

running · port 16395startup: 0.1s

🏗️

AI Appliances

Build and deploy self-contained AI systems as Docker images, bootable ISOs, or standalone binaries.

🤖

Coordinated Agencies

Orchestrate specialized AI agents into coordinated, multi-agent systems with fully defined control flow.

📝

Agents as Code

Declarative YAML configuration that is versioned, reviewed, and tested. No glue code, no boilerplate.

🛡️

Defined Control Flow

Predictable and auditable workflows. Inputs are declared, outputs are typed, and validations are enforced before any LLM is called.