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

# Projects Page

> Aggregate runs by project and preserve project memory

The **Projects** page (`/projects`) aggregates the runs of every experiment in one repository (whose project name is `name` in `starforge.yaml`): sweep variants, multi-round iterations, different stages — all in one place.

<Frame caption="Projects: one card per repo, run count and latest status.">
  <img src="https://mintcdn.com/starforge/GatXR2rI5-_Vm4_H/images/console/projects.png?fit=max&auto=format&n=GatXR2rI5-_Vm4_H&q=85&s=70a3fd8a89f289584b0ad0dd79987fe0" alt="StarForge projects page" width="2160" height="1350" data-path="images/console/projects.png" />
</Frame>

## Project view

* Each project shows run count, latest status, and best metric;
* Click into a project (`/projects/detail`): run list + key-metric comparison table + mini curve charts;
* Sweep variants automatically annotate grid axis values, so it's obvious which variant is which parameter set.

## Project memory

**Memory cards** inside a project preserve human conclusions: "lr 2e-6 + KL 0.05 is the most stable", "reward ceiling moved up after data v2". Written-down lessons travel with the project, so whoever takes over doesn't have to dig through chat history.

## Config comparison

Any two runs in a project can diff their config snapshots — "what differs between these two runs" without eyeballing YAML.

<Tip>
  Every submission of every experiment in one repository (named by `name` in `starforge.yaml`) lands in the same project; sweep variants are distinguished by `sweep_id`.
  Project names need not be globally unique — the grouping key is `@<username>/<project>`, so identically named projects from different users never collide.
</Tip>