Semogram Docs
Getting started

Read your first run

Read a specific pipeline run, inspect step diagnostics and produced artifacts, verify output evidence and choose a supported recovery or rerun path.

A run is the record of what a saved workflow actually executed. Use this page to check progress, find a failing step and verify the outputs from your first execution.

Steps

  1. Open the pipeline in Pipeline Studio and open its run history. Select the exact run, rather than assuming the latest run used your current draft.
  2. Check status and progress: queued, running, completed, failed, partial, paused, or canceled. Read the saved run context and individual step results.
  3. Inspect errors and connector diagnostics. Counts for fetched, emitted, skipped, failed, and duplicate records help distinguish source failures from mapping or downstream failures.
  4. Follow produced artifacts to their source references, upstream artifacts, and run/node linkage. Evidence coverage depends on the producing capability; verify it on the output you intend to use.
  5. If the run failed or stopped, inspect Recovery. Resume only when the recovery state allows it and you understand which completed work is retained. Correct structural problems in a new workflow version before a fresh run.
  6. Review relevant artifacts or assertions using their review controls. Artifact review supports approved, corrected, and rejected decisions with notes. A retry or resume is an execution action, not a review verdict.

Saving a plan version does not create a run. A completed run means execution finished; it does not prove the answer is correct or the destination has the intended business state. Verify both.

See Runs and Troubleshooting.

Keep a short execution record

Note the saved pipeline version, run identifier, source context and output you checked. If the execution failed, keep the first useful step diagnostic rather than only the final failed status. If it wrote externally, verify the receiving state before attempting recovery.

This record makes a later rerun comparable and gives a teammate enough context to investigate. See Troubleshooting if the next operation is unclear.