Documentation navigation
Getting started Quickstart Installation Core concepts Product tour
Building trees Creating nodes Prompt blocks Connectors Roles and governance Skills Co-pilot Collaboration
Run and inspect Execution Runs and traces Versions and rollback
SDK Python SDK Node.js SDK
Integrate API MCP and bundles
Providers AI connections Hosted providers for testing
Deploy Self-hosting Orchetree Cloud Environment variables

Runs and traces

Every tree execution creates a run record and node-level traces. Use the Runs area to test a system and determine where its behaviour came from.

Run record

The run view includes:

  • status, start time, and duration
  • original input and final output
  • total token and cost information when the provider reports it
  • event timeline and failure message
  • one trace for each attempted node

You can cancel an active run, repeat an earlier run, export run data as JSON, or request Co-pilot analysis.

Node traces

A trace records the node input, output, status, timing, token usage, errors, and relevant runtime metadata. Compare parent and child traces to determine whether a problem began in source data, a prompt, a connector, a model call, or output validation.

Live updates

GraphQL WebSocket subscriptions use the graphql-transport-ws protocol:

  • runStatus(runId) emits the run until it reaches a terminal state
  • runTraces(runId) emits trace updates during execution

The application also publishes AI run start and end events for global activity indicators.

Debugging order

  1. Confirm the run input and resolved version.
  2. Find the first failed or unexpected node trace.
  3. Inspect connector and skill context.
  4. Check the resolved AI connection and model.
  5. Compare the output with its contract.
  6. Change one prompt or policy at a time, then run again.

See Execution, Versions and rollback, and API.