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

# Review Graphs, Timelines, Files, and Diffs

> Understand the product review surfaces for research history.

Onyx turns reported experiments into reviewable project state.

## Campaign Graph

The graph shows the campaign, experiments, hypotheses, workers, and provenance links. Experiment nodes include metric context, git status, and result refs.

Use it to answer:

* Which directions exist?
* Which experiment is best?
* Which hypothesis or worker has gone stale?
* Which ideas inspired or superseded later attempts?

## Campaign Overview

The campaign review surface uses bounded overview, paginated experiment detail, and metric-series reads for stored experiment, hypothesis, worker, setup, summary, and metric projections. It does not perform hidden writes in read handlers; explicit reconcile actions refresh stale worker/hypothesis state and git status.

Use it to compare:

* succeeded vs failed attempts;
* improvements vs regressions;
* recent experiments vs earlier best results.

## File Tree

The file tree is a review interface, not a general code browser. It shows:

* files changed between the experiment result commit and base commit;
* files under `onyx/` for context;
* protected script markers.

## Diff View

Diffs are read from actual git commits. The default comparison uses the selected experiment and the previous best commit when available.

Use diffs to understand why a metric changed before building on a result.