What it does

AgentGrid puts several coding agents on one infinite canvas. You bring your own subscriptions to providers such as Claude, Codex, Cursor, OpenCode, Antigravity, Kimi, Grok, Devin and Pi, and run them as workers beside plans, notes and terminal output. A Coordinator chat on the right of the canvas can create agents, arrange the workspace and direct everything from one conversation. A master agent can plan, build and review by handing focused tasks to workers, each with its own chosen harness, and you can follow up in the same conversation. Each agent has a source control view of its worktrees, so you can read a diff beside the conversation. A workspace menu shows projects and agents, flags unread results, and shows provider usage and reset times. You can switch the model in any agent and keep the conversation, start a call from an agent pane, and reopen the app to resume saved projects and sessions. It runs on macOS, Windows and Linux. The page does not state AgentGrid's own pricing.

Capabilities

Infinite canvas for agents, notes and terminal output
Coordinator chat that creates and arranges agents
Master agent that delegates to workers, each with a chosen harness
Per-agent source control view of worktrees with diffs
Provider usage and reset times in one place
Switch models mid-conversation and resume saved sessions

When to use it

You run more than one coding agent and want them in a single workspace
You want to delegate planning, building and review across agents and read the diffs together
You already pay for AI subscriptions and want to use them from one place