StatefoldOne workspace.
Every agent you trust.
Bring your coding agents together with workflows you control, reusable skills, and one cloud workspace. From the first plan to the final review.
Explore the workspace
- Claude Code
- Codex
- OpenCode
- Pi
Gemini CLI
Workflows,
your way.
Give each stage the agent it needs. Choose the harness, model, instructions, and skills, then save the workflow for the next task. A quick fix or a full plan, build, and review: you set the sequence.
- 01
Plan
Claude CodeDefine the approachwritesplan.md - 02
Build
CodexImplement the planreadsplan.mdwritesbuild.md - 03
Review
Gemini CLIApproval gate: ask youreadsbuild.md
An example workflow. Each agent keeps its own native conversation; files carry the work forward.
Your stages. Your instructions.
Choose what each agent does, which skills it uses, and how independently it works.
Files carry the handoff.
Plans, code, and review notes stay with the work. The next agent starts from those files.
You decide what runs next.
Continue automatically, pause for approval, or loop back to an earlier stage with a repeat limit.
A memory
for every project.
Your project is more than its files. It’s the decisions you’ve made, the conventions you follow, and the things you’ve already figured out.
Project memory saves useful context from your work and brings relevant notes into later tasks. Less re-explaining. A better starting point.
Opt in by project. Inspect, edit, or forget saved notes. Native agent conversations stay separate.
Your skills,
wherever you work.
Keep personal skills in your global library, scope them to a project, or version them alongside your code. Choose the skills each stage needs.
Supported upstream sources update automatically. Existing tasks keep their pinned versions, so an update doesn’t change work already underway.
Steer now.
Queue what’s next.
A new constraint doesn’t have to mean starting over. Guide the active turn when the harness supports it, or line up a follow-up for when it finishes.
Keep the existing API unchanged.
Current turnThen add coverage for expired tokens.
Next turnQueued messages wait for the current turn to finish, even when steering is available. You choose when the next instruction enters the work.
Your connections.
A clearer next move.
Bring your provider accounts and API keys. See which connections have quota before choosing the agent for the next stage.
| Harness | Capacity |
|---|---|
| CodexChatGPT account | Window exhaustedResets in 24 min |
| OpenCodeProvider API key | AvailableSuggested for build |
| Claude CodeAnthropic API key | AvailableReady for review |
Illustrative availability. You see and approve the workflow before it runs.
When a connection reaches its limit, Statefold shows when it resets. Wait for that connection or choose another with quota available.
Choose each stage yourself, or review a proposed workflow with the reasoning behind every choice.
Start where
you’re working.
Send your current Git workspace to a cloud task, including uncommitted changes.
No cleanup commit. No push just to give an agent your latest files.
Your local branch and index stay untouched.
~/your-repo
$ statefold task new \
"Add password reset"Keep the conversation, files, and final diff together. Download saved work into a separate directory when you’re ready.
Local snapshot → Cloud workspace → Browser review
Questions
About agents, connections,
and taking your work to the cloud.
Is Statefold a coding agent?
No. Statefold is the workspace and workflow layer around native coding agents. You choose the harness, model, connection, and instructions for each stage.
Can I bring my own accounts?
Yes. Connect your own provider accounts and API keys. Subscription support depends on the provider and harness; Statefold does not promise that every subscription works with every agent.
What moves between agents?
Working files and explicit handoff documents, such as plans and review notes. Each agent keeps its own native conversation. Statefold sequences the stages, not the turns inside them.
Is project memory the same as a shared conversation?
No. Project memory keeps contextual notes for later tasks; it does not transfer a harness’s native conversation. Opt in by project, then inspect, edit, or forget saved notes.
Do skill updates change an existing task?
No. Supported upstream sources can update your skill library automatically, while existing tasks keep their pinned skill versions. Repository skills stay with your code.
Do I have to move my local work to a new branch?
No. The CLI snapshots your current Git workspace, including uncommitted work, without changing your local branch or index. Review cloud work in the browser and download the result separately.