Use Nerve day to day

You do not need to learn every panel before starting. Begin with one conversation in Supervised mode, then add the workflows you need.

The everyday loop

1. Describe the outcome. Use the composer to add a prompt, project paths, or an image.
2. Keep the authority appropriate. Choose Coding or Planning mode and a permission level.
3. Stay involved. Read the transcript and resolve approvals, questions, and plan reviews.
4. Inspect the result. Edit or review files, inspect the Git diff, and run focused validation.
5. Continue or branch. Steer active work, return to an earlier history point, or start another conversation without losing the original record.

Find the right guide

Learn the interfaceUnderstand projects, dock panels, center tabs, compact layouts, and the built-in tour.

Work with conversationsCreate and search conversations, read transcripts, branch history, and recover interrupted work.

Review code and filesEdit small text files, inspect context, use notes, and review Git or pull requests.

Delegate workDelegate research and implementation to sub-agents and define your own profiles.

Supervise long-running commandsSave launch configurations for servers and watchers, and inspect asynchronous Bash commands.

Customize NerveConfigure models, tools, skills, suggestions, appearance, notifications, storage, and system behavior.

Keep these concepts separate

• An agent performs work in a conversation. A lead can delegate to sub-agents created from profiles such as Explore and Developer.
• A to-do is the agent’s structured checklist for its current work.
• A launch configuration saves how to run a server or watcher; its runs have separate status and logs.
• Async Bash runs a conversation-owned shell command in the background and notifies the agent when it exits.
• A skill supplies instructions. Enabling one does not grant extra tool permissions.

If you have not installed Nerve or connected a provider yet, use the guided quickstart first.