Skip to main content

Evaluate OpenAgent

Use this page if you are trying to answer one question:

Is OpenAgent actually for me?

OpenAgent is a local-first workflow for people who already think in repos, docs, and graph context.

It is a strong fit if you want to:

  • work from Obsidian Canvas instead of a plain chat window
  • keep prompt context visible as nodes, files, and follow-ups
  • run Codex against a real local repo on disk
  • write results back into the graph instead of losing them in chat history

It is probably not a fit if you want:

  • a hosted SaaS product
  • browser-only usage
  • Windows or Linux support today
  • a workflow that does not depend on Obsidian Desktop and Codex Desktop

The Main Product Surface​

The primary OpenAgent workflow is:

  1. open a repo workspace in Obsidian
  2. select nodes on a Canvas
  3. start a Codex thread from that selection
  4. review progress in the side panel
  5. keep the final result attached to the graph

If that sounds like how you already think about work, OpenAgent will feel natural.

The Four Main User Groups​

OpenAgent docs now assume four main reader types:

1. Evaluator​

You want to understand the product before installing anything.

Read next:

2. Obsidian user​

You want the core product workflow inside Obsidian Canvas.

Read next:

3. Installer​

You already want it running and need the setup path.

Read next:

4. Contributor​

You want repo context, architecture, and the development workflow.

Read next:

Mobile Is A Separate Path​

Mobile is not just a smaller version of the Obsidian flow.

It is a separate product path where:

  • Convos is the chat surface
  • OpenAgent stays local on your Mac
  • Codex still runs against your local repo

If that is the workflow you care about, skip the Obsidian path and read:

If you are still deciding, read Use OpenAgent in Obsidian.

That page is the fastest way to understand the main user experience without dropping straight into developer detail.