Run your first task
1. Connect a model
Section titled “1. Connect a model”Open Settings → Models and connect at least one provider. Tellnova supports subscription providers, API-key providers, and custom OpenAI-compatible endpoints.
The green Shown in task composer status means the provider is ready to select for a task.
2. Choose a project
Section titled “2. Choose a project”Return to the task composer and choose either:
- a GitHub repository, which Tellnova can clone and update; or
- a local folder on this machine.
For a GitHub repository, sign in with an account that can read the repository. For a local folder, select the folder on the machine that will execute the task.
3. Choose a workspace mode
Section titled “3. Choose a workspace mode”Use the workspace icon in the project row to choose where the agent works:
- Current checkout runs directly in the selected local repository.
- New isolated worktree creates a separate branch and working directory.
The first task for a local repository defaults to its current checkout. Use an isolated worktree when you want parallel tasks or a pull-request review flow.
4. Describe the result
Section titled “4. Describe the result”Write the outcome you want, then select the provider, model, and reasoning level. Include constraints and a verification target when they matter.
Add password reset to the sign-in screen. Preserve the current visual style,add tests for the request flow, and run the existing authentication tests.5. Review and ship
Section titled “5. Review and ship”Tellnova streams the agent’s progress into the task. When it finishes:
- a current-checkout task can commit or commit and push its selected changes;
- an isolated-worktree task can push its branch and open a pull request.
Review changed files and test output before shipping the result.