Value-trace

Use when: you have a record and want to explain one cell — "where did this email come from?", "why did Enrich Person error?", "why didn't Push to HubSpot run?", "why is this value empty?". This walks a single column backward through its dependencies to find the origin or the root cause. Causation, not state — /tables-trace gives you the state snapshot first.

The method combines two things:

• Structure — column settings tell you the edges: who feeds whom ({{f_xxx}} tokens in formulaText / inputsBinding / formulaMap), and the run gate (conditionalRunFormulaText).
• This row's data — clay tables rows get tells you what each cell on this specific row actually holds and its status.

Structure gives the edges; the row gives the values along them. You need both.

{{f_xxx}} references never cross tables, so the entire walk stays within the row's own table.

1. Anchor to the row and the target cell

You need this row and the target column. If you don't have the row yet, locate it first (/tables-trace, or a single value search — tables query when the table is query-enabled, else rows list --filter). Identify the target column — the one whose value, error, or non-run you're explaining — and its f_id. The explanation needs the row's full cell data — every cell's status, value, fields, and error; a tables query result already carries all of that, and rows get provides it when you hold a rec_... id (step 3).

2. Get the structure — clay tables columns get