Your coding agents, together

Mains 0.5 is here — our first stable release. Give a coding agent a task, watch it work, and review what changed, all in one desktop app.

your agents, together

• Pick the agent for the job: Run GitHub Copilot, Claude Code, OpenAI Codex, or Cursor from the same place.
• Come back to a conversation: Resume a session later, keep it going, or fork it to try a different approach.
• Connect your tools: Add MCP servers so agents can work with the APIs and systems your team already uses.

a workspace for each task

• Keep changes separate: Each task gets a Git-backed workspace with its own diff, so you can see exactly what the agent changed.
• Start with the issue: Bring in tasks from GitHub, GitLab, Linear, Jira, Asana, or Trello. Add files, diffs, or Sentry errors when the agent needs more context.

see the work before you ship

• Stay in control: Approve tool requests as they happen, or choose trusted tools ahead of time.
• Review the result: Inspect the diff, request a code review, and look back at the files, logs, and cost of each run.