What it does
intentic is a free, MIT-licensed workspace that lets you direct several coding agents from a browser while the work runs in a sandbox on hardware you own. You hand out tasks, agents plan and work in parallel, and you review the results. A single board shows every agent, sorted by who needs your attention, with the model, branch, diff size and cost on each card. Every agent works on its own git worktree, so parallel agents never edit the same checkout. Agents read the code, write a plan and wait for your approval, and finished work waits on its branch until you read it file by file and land or discard it. Permission modes range from plan-only to fully automatic and can be set per turn. The sandbox is a Docker container on a laptop, desktop or server and keeps running after you close the browser, so any browser, a phone included, can reopen the same board. Agents use your own Claude, ChatGPT, Cursor, Google, Grok, Kimi or Z.ai accounts, or a local model, and the site says intentic does not meter tokens or add a markup. Agents can connect to services such as GitHub, PostgreSQL, Sentry and any MCP server, and can be started by events like a schedule, a push or a failed pipeline. The site calls the product new and advises starting with low-risk work.
Capabilities
- Parallel coding agents, each on its own git worktree
- Plan approval before any code changes
- File-by-file diff review before work lands
- Keeps running after the browser is closed, reachable from any browser including a phone
- Works with your existing AI accounts or a local model
- Event triggers such as schedules, pushes and failed pipelines
When to use it
- You want several coding agents working on one repository without conflicting edits
- You want agents to run on hardware you own and keep working while your laptop sleeps
- You already pay for AI coding subscriptions and want one place to manage them
