Declarative YAML Specs

Instead of writing Rust code, you can define runs as YAML files and execute them directly with the CLI.

1. Supported kinds

The runtime currently supports four top-level spec kinds:

  • agent — a single agent run inside one isolated VM
  • pipeline — multiple agent boxes composed into explicit stages
  • workflow — command-oriented workflow steps executed in a shared sandbox
  • sandbox — a bare VM definition used for interactive or shell-oriented flows

2. Agent spec

A single agent spec declares the sandbox, LLM provider, prompt, and skills:

api_version: v1
kind: agent
name: hn_researcher

sandbox:
  mode: auto
  memory_mb: 1024
  vcpus: 1
  network: true

llm:
  provider: claude

agent:
  prompt: >
    Analyze current HackerNews top stories and produce
    a tactical briefing for AI engineering teams.
  skills:
    - "file:examples/hackernews/skills/hackernews-api.md"
  timeout_secs: 600

3. Pipeline spec

A pipeline spec chains multiple boxes with explicit stage ordering:

api_version: v1
kind: pipeline
name: quick-pipeline

sandbox:
  mode: auto
  memory_mb: 512
  network: true

pipeline:
  boxes:
    - name: analyst
      prompt: Extract key facts from the input.
      skills: ["agent:claude-code"]
      timeout_secs: 300

    - name: writer
      prompt: Turn facts into a concise executive summary.
      skills: ["agent:claude-code"]
      timeout_secs: 300

  stages:
    - type: box
      name: analyst
    - type: box
      name: writer

4. Running with the CLI

From a clone of void-box, use voidbox run --file to execute any spec file:

voidbox run --file examples/hackernews/hackernews_agent.yaml
voidbox run --file examples/specs/pipeline.yaml

5. Environment overrides

Override the LLM provider at runtime without changing the spec:

VOIDBOX_LLM_PROVIDER=ollama VOIDBOX_LLM_MODEL=phi4-mini \
  voidbox run --file examples/specs/pipeline.yaml

Environment variables take precedence over spec-level llm.provider settings.

6. Next

Learn how to compose pipelines in Rust, or set up local LLMs with Ollama.