If you're building with AI agents — Claude, Gemini, Vertex, or anything CLI-based — you've probably felt the pain: terminal tabs multiplying, no easy way to compare agent outputs side-by-side, losing scroll history when you close a window, and zero visibility into what things cost.
I built Agenv to fix that.
What is Agenv?
Agenv (The Agent Development Environment) is a full web-based IDE purpose-built for running and monitoring AI agents. Think of it as a workspace where you can:
- Run multiple agents side-by-side in split terminal panes
- Edit code with syntax highlighting for 30+ languages
- Stage, diff, and commit with built-in git integration
- Track token usage and costs in real-time per session
- Persist sessions that survive browser crashes and reconnects
It runs as a desktop app (via Electron) or a web server you can access from any device — including your phone.
adibenmati
/
agenv
The agent development environment — build, run, and monitor AI agents from a full web IDE with split terminals, file editing, and session management
Agenv
The agent development environment. Build, run, and monitor AI agents from a full web-based IDE with split terminals, integrated file editing, git management, and persistent session control.
Agenv gives you a complete workspace for agent development — not just a terminal, but an environment where you can run multiple agents simultaneously, inspect their output in split panes, edit code with syntax highlighting, track costs, and manage everything from any browser or the desktop app.
Video.Project.mp4
Install
npm install -g @adibenmatdev/agenv
Or run without installing:
npx @adibenmatdev/agenv
For the desktop app experience, also install Electron (optional):
npm install -g electron
Quick start
# Launch the environment (desktop app if Electron is installed, web mode otherwise) agenv # Force web mode (opens in browser, accessible from any device) agenv --web # Web mode on a custom port agenv --web --port 8080 # Launch directly into an agent agenv run claude --model
…
Install in 10 seconds
npm install -g @adibenmatdev/agenv
Or try it without installing:
npx @adibenmatdev/agenv
Then just run:
agenv
That's it. If Electron is installed globally, it opens as a native desktop app. Otherwise it starts a web server and opens your browser.
The Features That Matter
Split Terminal Panes
Run multiple agents simultaneously. Each gets its own persistent PTY session. Switch between them with Alt+1-9 or the tab bar. Drag-and-drop files directly into terminals.
Integrated Code Editor
Full file editor with syntax highlighting, file explorer with tree view, quick file search (Ctrl+P), and text search across your project (Ctrl+T). No more switching between your terminal and your editor.
Git — Without Leaving the IDE
Stage files, view diffs, write commits, and push — all from the sidebar. See changed files at a glance, discard or unstage selectively. It's not a replacement for the CLI, but it covers 90% of what you need mid-session.
Session Persistence
Close your browser tab. Shut your laptop. Come back later. Your sessions are still there — scrollback history, running processes, everything. Sessions are encrypted at rest with AES-256-GCM and stored locally.
Cost Tracking
Real-time token usage and cost monitoring per agent session. Status indicators show whether each agent is running, waiting, or errored. Know exactly what your experiments cost before the bill surprises you.
Desktop or Web — Your Choice
# Desktop app (default when Electron is available)
agenv
# Web mode — access from any device on your network
agenv --web --host 0.0.0.0
# Need remote access? Built-in ngrok support
agenv --web
# then use the ngrok integration from the UI
Security by Default
- Desktop mode: no open ports, no network exposure
- Web mode: 128-bit random token auth, scrypt-hashed passwords, rate limiting, CSP headers
- All persistent data encrypted with AES-256-GCM
- Electron sandbox enforced
CLI at a Glance
agenv # Launch (desktop or web)
agenv --web # Force web mode
agenv --web --port 3000 # Custom port
agenv run "claude" # Start with a command running
agenv stop # Stop the server
agenv set agent gemini # Configure default agent
Only 4 Runtime Dependencies
I deliberately kept Agenv lean:
-
express— HTTP server -
ws— WebSocket -
@lydell/node-pty— terminal emulation -
qrcode-terminal— QR code for mobile access
The entire frontend is vanilla JavaScript modules — no React, no Vue, no build step. The UI is fast because there's nothing to get in the way.
Why I Built This
I was spending most of my time running AI agents in terminals — tweaking prompts, comparing outputs, iterating on code. But the tooling around it was terrible. I had 15 terminal tabs open, couldn't find anything, lost scroll history constantly, and had no idea what my API usage looked like.
Agenv is the environment I wanted. One place where I can run agents, read their output, edit code, and track costs — without juggling windows or losing context.
Try It
npx @adibenmatdev/agenv
GitHub: github.com/adibenmati/agenv
npm: @adibenmatdev/agenv
License: Apache-2.0
If you're building with AI agents, give it a spin. Issues, PRs, and stars are all welcome.