Skip to main content

Examples

Working YAMLs. Every example here is verified to compile and run against a live daemon. The full collection (with deeper context) lives in language/15-examples.md; this section is the curated subset that backs the doc's recipes.

Smallest possible app​

The minimum that compiles and runs:

app:
  app_id: smallest
  name: Smallest

agents:
  - id: a
    role: assistant
    brain:
      provider: ollama
      model: qwen25-7b-gpu:latest
      backend: openai_compat
      config:
        base_url: http://localhost:11434/v1
        api_key: ollama
    system_prompt: "Reply with exactly one word: pong."

Deploy + chat:

digitorn dev deploy -d http://127.0.0.1:8000 smallest.yaml
digitorn dev chat smallest -m "ping"
# → "pong"

Canonical 8-block reference​

The example documented in Language - Complete example exercises every top-level block. Use it as a template when authoring a new YAML.

Real-world examples​

For production-shape apps (multi-agent, RAG, channels, sandboxed MCP, credentialed brains), see language/15-examples.md. Each example there names the modules it requires, the credentials it expects, and the deploy command.

How examples are verified​

Every YAML in the documentation is verified by the same automated flow:

  1. Save the YAML to a temp file.
  2. digitorn dev deploy -d <daemon> <file> - the compile + bootstrap path. A failure here means the YAML is structurally wrong.
  3. client.send_live(session, "<prompt>") from the Python testing SDK - opens a Socket.IO stream and waits for message_done.
  4. Assert against result.content and the event sequence.

A YAML that doesn't pass step 3 doesn't ship in the docs.