Agentic Development Environment

Runova

Run the work, not just watch it.

Put Claude Code, Codex, Cursor, Gemini or a plain shell in panes side by side. Git, diffs, tests and pipelines sit right next to them, so a change goes from prompt to review to commit without leaving the app.

Download Runova runovahq.com
  • Free and open source (MIT)
  • Windows, macOS and Linux
  • Built by SmartClouds
The Runova workspace: an agent in one pane, a diff open in the next, and the git panel beside both

Works with the coding CLIs you already use

  • Claude Code
  • Codex
  • Cursor
  • Gemini
  • Grok
  • OpenCode
  • GitHub Copilot CLI
  • Qwen
  • Aider
  • Goose
  • Cline
  • and a plain shell
Download

Get Runova for your machine

Version 3.2.0. Free and open source under the MIT license. Every link points at the newest release.

Windows Installer, x64 EXE macOS Apple silicon DMG macOS Intel DMG Linux AppImage, x64 AppImage Debian & Ubuntu Package, x64 DEB Android Companion app APK
On Windows, with winget winget install SmartClouds.Runova
All builds & checksums Source on GitHub
Why an ADE

Built for agents writing the code and you steering

An IDE is built around a person typing code. Runova is built around the agents doing the typing, so it has to do a few things an editor never needed to.

  1. Several agents in view at once A workspace is a tree of panes, each with a CLI or a shell in its own folder. Runova shows which agent is working, which is waiting on you and which is done.
  2. The loop closes in one window Read the diff, stage hunks, run the tests, send the failure back, commit, tag and watch the pipeline, all next to the agent that wrote the code.
  3. Tokens and quota up front Cost and rate limits stay on screen, and an agent tab can wait out a limit and carry on by itself.
  4. Long jobs outgrow one session Relay cuts the work into tasks that fit an agent's context and hands what each session learned to the next one in a file.
  5. Agents read and run things So skills get scanned before an agent reads them, secrets live in an encrypted vault, and risky SSH commands stop for your approval.
Inside the app

Take the tour

Every screen below is the real app with demo data, in the default Dark theme.

Relay: A task board where every card fits one agent session. Cards move from To do to Done on their own as each session hands off to the next.
Relay A task board where every card fits one agent session. Cards move from To do to Done on their own as each session hands off to the next.
Token usage: Tokens, cost, quotas and spend per project, across more than 60 providers.
Token usage Tokens, cost, quotas and spend per project, across more than 60 providers.
Projects: One card per repo, with its agent, tags and a Run button for its dev server.
Projects One card per repo, with its agent, tags and a Run button for its dev server.
Monitoring: What is up, what is failing, and how each environment has been doing.
Monitoring What is up, what is failing, and how each environment has been doing.
Deploy: Your own Linux servers over SSH: apps, containers, sites, certificates and the firewall.
Deploy Your own Linux servers over SSH: apps, containers, sites, certificates and the firewall.
AI CLI Manager: Detect, install and update every coding CLI, and save the model and flags each one launches with.
AI CLI Manager Detect, install and update every coding CLI, and save the model and flags each one launches with.
Dashboard: System health, network, GitHub activity and token usage at a glance.
Dashboard System health, network, GitHub activity and token usage at a glance.
Themes: The same workspace in Dark, VS 2026, VS Code Dark and Light.
Themes The same workspace in Dark, VS 2026, VS Code Dark and Light.
The core

Seven things the whole app is built around

Multi-pane workspace

Split the window as far as you like and drop a CLI, a shell, a file, a diff or a browser into each pane. Layouts save per project, and a git worktree gives each task its own branch and folder, so agents never touch each other's files.

Agent status

Every agent tab reports whether it is working, waiting on you or finished. Click a status to jump straight to that pane.

Git, diffs and the editor

Stage by hunk, commit, branch, stash and open pull requests from the side panel. When a merge stops on conflicts, hand them to an agent with Fix with AI.

Tests

Finds Vitest, Jest, Playwright, pytest, Go, cargo, .NET, Flutter and more. Results stream in live, and failures go to an agent in one click.

Review

Multi-agent code review for bugs, security, performance and consistency, over the working tree, a branch or the last few commits.

Relay

Split a big job into task files that each fit one session. Every session leaves a handoff with what was done, what was decided and the next step, and the next session reads it first. With Claude Code, Runova watches real context use and calls the handoff at 80%.

Token usage & auto-continue

Tokens, cost and limits across 60+ providers, read from local logs for Claude Code and Codex with no API key. Auto-continue types "continue" for you once a limit resets.

Everything else

The tools around the work, so you never leave the app

The menu is grouped the way the work goes: build it, run it through agents, ship it, and connect to everything around it.

Build

Turn an idea into a plan and a prompt, then test the API it talks to.

  • Blueprint
  • Prompt Builder
  • Prompt history
  • Scheduled prompts
  • Ask AI
  • API Client

Agents

Install and update the CLIs, plus the skills and MCP servers they use.

  • AI CLI Manager
  • Launch defaults
  • Skill Marketplace
  • Skill security
  • MCP servers
  • Agent Tools

Ship

Run workflows, deploy to your own Linux servers and watch what is live.

  • Pipelines
  • Deploy
  • Monitoring
  • Sentry errors
  • Docker
  • Android

Connect

Reach other machines and keep every credential in one encrypted place.

  • Remote & mobile app
  • SSH & SFTP
  • Remote Desktop
  • Vault
  • Telegram alerts
The AI Pet showing a token report

And a pet on your desktop

An optional companion that lives outside the window. Click it for a token report, and it tells you when a pipeline fails, the network drops or an agent changes status. Bring your own GIF, PNG or WebP character if you like.

Security

Agents act on what they read, so Runova checks first

Skill scan

A static check over 14 risk categories before an agent reads a skill, with the file, the line and a 0 to 100 score. An installed agent can add a second opinion, but it can only pull the verdict down, never up.

Encrypted vault

Passwords, API keys, SSH logins and environment files, encrypted on your computer behind a master password, with auto-lock and a clipboard that clears itself.

Approval for risky commands

When an agent runs commands on a server over SSH, anything risky stops and waits for you to say yes.

What the skill scan looks for

  • Prompt injection
  • Data exfiltration
  • Credential theft
  • Privilege escalation
  • Supply chain
  • Remote code execution
  • Anti-refusal
  • System prompt leakage
  • Memory poisoning
  • Unsafe output handling
  • Dark-pattern payments
  • Hidden content
  • Destructive actions
  • Overbroad permissions
The name

A runway and a new star

A runway is where launches happen, and a nova is a new star lighting up. You bring an idea, agents build it with you, and it leaves as something shipped. The logo says the same thing: an R drawn as a runway, with a star taking off from its corner.

Formerly AgentMate Updating moves your settings, projects, secrets and terminal sessions over on its own. On a phone, install the new Runova app and pair it again. WordPress sites need the Runova Connector in place of the old one.
runovahq.com

The full tour, every build and the news feed live on the Runova site

Read about every feature in depth, grab any build with its checksum, and follow the technology news that also feeds the dashboard inside the app.

Visit runovahq.com Star on GitHub

Rolling Runova out to a team?

Tell us which coding agents your team uses today. We will help you set up workspaces, Relay and token tracking so everyone is productive from day one.

Talk to us