Engine-neutral work
Use Claude Code, Codex, Cursor, or another connected engine. Switch per task without losing the project record.
AI coding agent workspace
Build software with AI coding agents in one product workspace.
Iterel is a desktop-first product-creation studio for builders and product teams already using Claude Code, Codex, Cursor, or other coding agents.
Turn requests into tracked tasks, give every engine the right project context, run parallel work in isolated worktrees, inspect the exact changes, and keep Design, review, and collaboration tied to the product.
Why Iterel
Coding agents move fast. Product context fragments faster.
Agent transcripts, branches, screenshots, task trackers, design files, and review threads each hold a different part of the work. The engine may finish a change while the team still cannot see the intent, evidence, or next decision in one place.
Iterel keeps that surrounding product record durable. The engine still writes code; Iterel makes the work understandable, reviewable, and continuable.
The working loop
From request to reviewed change.
- 01
Connect the project and engine
Open a repository, local folder, or blank project, then use the coding engine you already trust.
- 02
Give the task its context
Attach the objective, selected screen, files, specs, decisions, and project rules that should govern the next round.
- 03
Run work in an isolated boundary
Managed writing tasks use separate worktrees. Dependencies and likely territory overlap stay visible before work moves forward.
- 04
Inspect the evidence
Review the conversation, exact Changes, checks, activity, and task-only Preview from that task’s worktree.
- 05
Accept, discard, or follow up
Keep the decision and receipt attached to the task, then start the next round without rebuilding its context.
Visual coding agent UI
A visual interface above the terminal session.
If you are looking for a GUI, dashboard, task manager, or command center for coding agents, Iterel adds the product record that an individual engine interface does not own.
Persistent product context
Carry accepted decisions, constraints, repository state, design intent, task history, and failures between rounds.
Visual product work
Build and Design live in one studio, with tracked Design delegation and an explicit reconciliation path back to code.
Review before acceptance
Changes, verification, ownership, engine, and round receipts stay visible at the exact task boundary.
Product-team collaboration
Members share project state and comments. Reviewers can inspect one saved target without receiving project or machine access.
Local depth, cloud continuity
Keep unlimited local projects on your machine. Create a collaborative cloud project only when sync and live collaboration are needed.
Clear responsibilities
Keep the tools you use. Give each one a clear job.
- Your coding engine
- Executes the task with its own model, subscription, and supported controls.
- Your IDE and Git
- Remain the direct code-editing and version-control foundations.
- Iterel
- Owns the tracked task, context, isolated run boundary, product view, receipt, and review loop.
Questions, answered
What an AI coding agent workspace changes.
Does Iterel replace Claude Code, Codex, Cursor, or my IDE?
No. Your coding engine executes the work and your IDE remains available for direct code editing. Iterel adds the shared task, context, design, review, and collaboration layer around that workflow.
Can I manage multiple coding agents in one visual workspace?
Yes. Iterel can organize tasks across supported connected engines, keep each managed writing task in an isolated worktree, and bring progress, needs-input, Changes, checks, and review back into shared project views.
Is Iterel a UI for Kimi Code?
Not a direct Kimi Code client. Kimi Code has its own terminal, web, VS Code, and ACP interfaces. Iterel can instead use a Moonshot API key as an experimental Kimi model provider routed through an installed compatible Claude Code engine, then add tracked tasks, project context, Design, review, and collaboration. Moonshot bills that model usage.
How does Iterel run several coding agents safely?
Iterel-managed writing tasks reserve separate worktrees. Explicit dependencies and advisory territory warnings make likely overlap visible, while every task keeps its own Changes and review boundary. Iterel does not claim that this prevents every semantic conflict.
Does Iterel upload my local project?
A machine-local project keeps normal Build and Design persistence on your computer, and opening Design does not upload it. Cloud collaboration and exact-target review are explicit actions with bounded data projections.
Can product managers and designers review agent work?
Yes. Signed-in project members can collaborate according to their role, and exact-target review links let viewers and commenters review a saved artifact, Design target, or Whiteboard target without project or machine authority.
Can I control a coding task from a browser or phone?
Paid remote control can continue eligible work through the same user’s online Mac. The phone is a companion for projects, tasks, needs-input, comments, Chats, command-centre attention, and Assets—not a mobile Design, Preview, or Terminal surface.
What is free?
Free includes unlimited machine-local projects with Build and Design, one editable collaborative cloud project, one connected engine, and unlimited exact-target review links, viewers, commenters, and comments. Your engine provider may still require its own subscription.
Start with your own engine
Keep the speed. Recover the product workflow.
Free includes unlimited machine-local projects with Build and Design, plus one editable collaborative cloud project.