Skip to content

First Run with CMD

Complete installation, initialization, and Workflow installation first. This walkthrough uses intent.

Prepare service execution

Add a real provider/model configuration to the generated intloom.yaml, retaining its existing fields and installed Workflow declarations:

llms:
  default:
    provider: openai-compatible
    model: your-model
    secret: ENV.MODEL_API_KEY
    baseURL: https://your-provider.example/v1

Replace placeholders and provide MODEL_API_KEY in the environment that starts the Service. Model configuration explains providers and roles.

intloom stop
intloom start
intloom doctor --execution cli
intloom looms
intloom run intent --prompt "Build a Todo app with local persistence"

Retain the returned Run ID. doctor checks readiness without calling a model; the Run receipt admits work but does not prove completion.

Continue the same Run

intloom wait RUN-example
intloom attach RUN-example --json

Replace RUN-example with the actual ID. When pendingAction appears, read its complete questions or confirmation context. Save the user's answer as JSON:

{"isConfirmed": true}

For a confirmation action, submit the file using the exact current action ID:

intloom answer RUN-example --action ASK-example --answer-file ./answer.json
intloom wait RUN-example

Question actions require an answer array instead; see Run management. Repeat observation and answering until terminal.

Read the result

intloom artifacts --loom intent
intloom records --loom intent
intloom artifact --loom intent --stage validation

Read the Validation findings and check results. Follow CMD usage for scripting and the Intent walkthrough for the four stages.