Multi-agent coding
Keep several coding agents in view.
Give each agent a session and a checkout. Follow progress, review code changes and test the running app from one browser workspace.
The working model
Choose how to run each agent
Run command-line agents in Terminal sessions. Use Agent Controller for supported agents when you want a conversation view with tools, diffs and approvals. Keep both beside your tests and dev servers.
Separate the work
Give concurrent writers separate branches or checkouts. Keep read-only analysis together only when file ownership cannot collide.
Choose a session type
Use Terminal for command-line tools or Agent Controller for a supported agent.
Check progress
Use session activity, attention, Git state, terminal output and browser evidence to see what is actually blocked or complete.
Review and merge
Review each agent’s changes, merge the work you want to keep, then close finished sessions and remove temporary checkouts.
What tlbx adds
Tools for following parallel work
- Terminal
- Run command-line tools with terminal output retained on the host.
- Agent Controller
- Read conversations, inspect tools and code changes, and answer approval requests for supported agents.
- Control
- Use commands or APIs to select sessions, send prompts, read output and check the app preview.
- Shared context
- Keep files, Git status and app previews beside the sessions doing the work.
Good fit
Independent features, code reviews, long builds and browser checks that can run in parallel.
Keep edits separate
Agents editing the same files can overwrite each other. Give concurrent writers separate checkouts and review changes before merging.
You decide what ships
Use the shared view to review changes and approvals. You remain responsible for merging work and approving destructive actions.
Common questions
Multi-agent coding FAQ
Does tlbx call the model provider?
Your agent runs locally with its own provider connection. Its authentication, data handling and model charges follow your provider configuration.
Can different agents run at the same time?
Yes. Each runs in its own session. Give agents that edit concurrently separate checkouts to prevent file collisions.
Can an agent control the browser?
Yes, when explicitly authorized. tlbx exposes scoped browser status, navigation, query, click, text and diagnostic helpers.
Does tlbx replace Git worktrees?
Use Git worktrees to keep concurrent edits separate. tlbx lets you follow the sessions and repositories working in them.