Discovered through heavy dogfooding. To see a node's complete result (e.g. a long codex report or a review), I repeatedly had to parse .odw/runs/<id>/events.jsonl by hand — the HTML report shows the prompt (truncated) and status/tokens, and runs show tails the journal, but neither gives one node's full returned value at a glance.
Proposed: odw runs show <id> --node <label> prints that node's full prompt + result, and/or the HTML report makes a node's result expandable (it already stores it). No new runtime behavior — just surfacing data that's already captured.
Severity: usability / nice-to-have.
Discovered through heavy dogfooding. To see a node's complete result (e.g. a long codex report or a review), I repeatedly had to parse
.odw/runs/<id>/events.jsonlby hand — the HTML report shows the prompt (truncated) and status/tokens, andruns showtails the journal, but neither gives one node's full returned value at a glance.Proposed:
odw runs show <id> --node <label>prints that node's full prompt + result, and/or the HTML report makes a node's result expandable (it already stores it). No new runtime behavior — just surfacing data that's already captured.Severity: usability / nice-to-have.