Skip to main content
Use this page as the quick index when something in Smithers feels wrong and you do not yet know which subsystem is responsible.

Start With the Symptom

Common Failure Modes

A task did not rerun

Smithers resumes by runId. If a node already completed for the current runId, it will be skipped on resume. Use a new runId for a fresh run, or inspect the persisted attempts with the CLI before assuming the scheduler is stuck.

An MDX prompt rendered as [object Object]

That usually means the MDX preload was not registered. Confirm that your preload script calls mdxPlugin() and that bunfig.toml points at the preload file.

The workflow code changed but the running task did not

Hot reload only affects work that has not been scheduled yet. In-flight tasks continue with the code and prompt they started with.

When in Doubt

  1. Check smithers status for the current run.
  2. Inspect recent frames with smithers frames.
  3. Query the SQLite database if the CLI summary is not enough.
Those steps usually tell you whether the issue is scheduling, agent execution, schema validation, or environment setup.