Skip to main content
Use this page when you want the shortest path to a working Fabriqa setup.

​
Before you begin

  • Make sure the Fabriqa desktop app is open.
  • Make sure the repository or folder you want to work on is available on the machine Fabriqa can access.
  • If you plan to use an agent provider, install or sign in to that agent first.
  • If you plan to use an LLM provider, make sure its credentials are configured in Settings.
1

Open a workspace

Pick or open the repository you want to work in. Fabriqa treats that repository as a workspace.If you are on the welcome screen, choose your workspace there. If you are already inside Fabriqa, use Open Workspace from the command palette.
2

Check your provider

Open Settings if you need to:
  • enable an agent provider
  • sign in to an agent provider
  • choose a default provider and model
  • confirm that your API-backed LLM provider is configured
If you want the full agentic workflow, prefer an agent provider. LLM providers are better for lighter chat flows unless the product surface says otherwise.
3

Start a chat

Create a New Chat in your current workspace. For isolated implementation work, create New Chat in Worktree instead.Choose a provider from the chat input toolbar if the default is not the one you want.
4

Send your first useful prompt

Try a prompt that produces visible workspace output, for example:
Inspect this workspace, summarize the current state, and suggest the safest next step.
Then try a second prompt that uses the workspace tools:
Find the main entry point, explain the current flow, and list the files I should review next.
5

Review the result

Use the surrounding surfaces as needed:
  • Reader for files the chat opens
  • Git Changes for file diffs
  • Terminal for commands
  • Timeline for step-by-step activity
  • Usage for token and cost details

​
Your first two workflows

​
Main session flow

Use a main session when you want to stay in the primary workspace folder and iterate quickly.

​
Worktree session flow

Use a worktree session when you want isolation for a feature branch, experiment, or task-specific implementation.

​
What to learn next