> ## Documentation Index
> Fetch the complete documentation index at: https://docs.droyd.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Compare and record analysis

> Compare evidence across experiments and preserve the reason for each decision.

Use the workflow history to compare your baseline, branches, scores, and
submissions:

```bash theme={null}
droyd workflows history <workflow-id> --json
droyd analysis add --title "Recall bottleneck" --text "The narrower query improved precision but reduced recall." --json
```

`latest` refers to the newest relevant remote state and `best` refers to the
best selected node, while `current` is only your local workspace pointer. State
the dataset, metric, and trade-off behind an improvement so the conclusion can
survive future score changes.

Optimize for transferable behavior, not guesses about private data or a single
leaderboard snapshot.