01
Separate projects and concerns with Spaces.
Keep unrelated agent sessions out of each other's way. Switch from a product push to a research thread without dragging every old tab along.
Native macOS workspace for CLI coding agents
Steer your AI coding agents
with precise feedback.
Rally runs Codex, Claude Code, and Gemini CLI in one native workspace — so you can review agent output, mark exactly what needs to change, and send it back as concrete instructions.
Opening in small cohorts — shaped by the people using it. Rough edges expected. Tell us what breaks →
Works with the CLIs you already use
Claude Code
·
Codex
·
Gemini CLI
Bring your own CLIs and subscriptions. Rally wraps the agents you already run.
Why it feels different
Not another terminal.
A control surface for agent work.
Rally is strongest when the work stops being one long transcript. Spaces, skills, feedback bundles, and usage signals keep agent sessions organized enough to steer.
02
Manage Skills and MCP from one place.
See the local tools your agents can use, turn servers on or off, and keep messy agent configuration visible instead of buried across dotfiles.
03
Collect many questions before sending one turn.
Mark several spans, add notes while reading, then send the whole review as a single instruction bundle. The agent gets the full shape of your feedback at once.
04
Watch token usage like an engineering signal.
Compare usage by engine, project, model, and session so heavy agent workflows stay observable instead of feeling like a surprise bill.
Why Rally
The human feedback loop is still copy-paste.
AI agents are good at producing work. Steering it is the hard part — you copy fragments, rewrite prompts by hand, and lose the thread between turns. Rally makes that loop explicit.
Without Rally
- Vague follow-ups: “fix the third paragraph”.
- Copy-pasted corrections, out of context.
- Lost intent between agent turns.
With Rally
- Select the exact span of output.
- Attach a note to that precise range.
- Send one deterministic instruction bundle.
The Input Harness
Turn review notes into instructions.
Rally ties feedback to the agent output it came from. The next turn gets specific context instead of another vague prompt.
- 01
Select
Highlight the exact span of agent output that needs attention.
- 02
Annotate
Attach a focused note to that selected range.
- 03
Collect
Gather multiple notes across one response.
- 04
Send
Deliver them as a single instruction bundle into the next turn.
More workspace details
The supporting surfaces stay close.
Multi-agent workspace
Run Codex, Claude Code, and Gemini CLI side by side, each in its own session.
Tabs & splits
Arrange panels for review, preview, terminal work, and long-running sessions.
Embedded terminal
A real shell — xterm.js and node-pty — right in the workspace.
Browser & doc readers
Open local previews, links, and Markdown/HTML files in a side pane.
Switch models mid-session
Change the model without losing the session or the transcript.
Local-first
No hosted Rally service, no prompt proxy, no model access.
How it works
Bring the CLIs you already use.
- Sign in to your CLIs.Use Codex, Claude Code, and Gemini however you already run them locally.
- Run & review in one place.Sessions, terminals, browser previews, and docs stay together.
- Annotate & send.Mark what matters and ship precise feedback into the next turn.
Local-first & transparent
Your keys, your CLIs, your files.
Rally runs entirely on your machine and launches the CLIs installed there. No hosted service. No prompt proxy. No model access.
Coding agents can run commands and edit files based on their permission mode — use trusted repositories and review permissions carefully.
Private Beta
Rally for macOS
Enter the access code from your invitation to unlock the private beta.
Access confirmed. Complete your details to download.
You're in. Your download should start automatically.
Didn't start? Download Rally again →
Installing the beta
- Open the downloaded DMG.
- Drag Rally to Applications.
- Open Rally from Applications and choose Open if macOS shows its standard downloaded-app confirmation.
Rally beta is signed and notarized by Apple, so the unidentified-developer bypass is not required. Future updates install from inside Rally.
Hit a rough edge? Send feedback → It shapes what ships next.
FAQ
Questions
Which agents does Rally support?
Claude Code, Codex, and Gemini CLI — whichever you already have installed and signed in.
Does Rally provide the AI models?
No. Rally wraps the CLIs on your machine. You bring your own subscriptions and keys; Rally never proxies prompts through a hosted service.
Where does my data go?
Everything runs locally. Rally launches your CLIs, reads the files you point it at, and keeps sessions on your machine.
How do I join the private beta?
Access is currently invitation-only. If you received a Rally access code, enter it in the download section above.
Is it free?
Rally is free during the private beta. You only pay for the agent CLIs you already use.
Is Rally signed and notarized?
Yes. Private-beta builds are signed with Rally's Developer ID and notarized by Apple. macOS may show its standard confirmation for an app downloaded from the internet, but you should not need the unidentified-developer bypass.
Windows or Linux?
Rally is macOS (Apple Silicon) for now.
Feedback
Shape the beta.
Rally is early. Bugs, rough edges, missing features — tell us and it goes straight to the maker. Email is optional; leave it if you'd like a reply.
Thanks — got it. This is exactly how the beta gets better.