agent-deckUnify AI Coding Assistant Sessions in TUI

If you're anything like me, your daily coding routine involves a constant dance between different AI assistants. Maybe Claude handles the core logic, Gemini steps in for optimization, and OpenCode helps with completions. This often leads to a chaotic mess of terminal tabs, forgotten session windows, and wasted time just trying to keep track of everything.

This is precisely the pain point agent-deck aims to solve. It's an open-source terminal session manager built from the ground up for AI coding agent scenarios. In essence, it provides a unified TUI (Terminal User Interface) that lets you manage conversations with multiple AI assistants from a single interface. It might sound like just another 'glue' tool, but in practice, it genuinely streamlines the workflow and cuts down on friction.

From Chaos to Control: The Genesis of agent-deck

The project, initiated by asheshgoplani and written in Go language, has already garnered over 2600 stars on GitHub. Its core idea is straightforward: with so many AI coding tools at our disposal, why not create a central control panel to manage all their sessions?

You might wonder, what's so special about a TUI? Don't we already have terminal multiplexers like tmux or screen? The key difference is that agent-deck isn't just about displaying multiple terminals side-by-side. It's purpose-built for AI conversation management, offering features like session summaries, quick switching, and historical review. It consolidates all your AI agent interactions into a single session list, eliminating the need to hunt through various windows.

Installation and First Impressions: Surprisingly Simple

As a Go project, the installation is remarkably clean. If you have a Go environment set up, a single command is all it takes:

go install github.com/asheshgoplani/agent-deck@latest

For those without a Go environment, the Releases page offers pre-compiled binaries; just download, extract, and run. Upon launch, you're greeted with a clean, two-pane TUI: the left side lists your sessions, and the right displays the current conversation. Keyboard shortcuts are intuitive: Tab to switch focus, Ctrl+N for a new session, and Ctrl+Q to exit.

• Supports specifying multiple AI agent API endpoints via a JSON configuration file.
• Sessions are automatically saved locally and can be restored after restarting.
• Allows custom system prompt templates for consistent instructions.

Initial setup requires configuring API keys for each agent. The documentation is clear, and you can be up and running in about five minutes. For developers comfortable with the terminal, the learning curve is practically non-existent.

Which AI Agents Does It Support? Beyond Just Claude

According to the project description, agent-deck currently supports Claude, Gemini, OpenCode (an open-source alternative to GitHub Copilot), Codex, and any model that provides an OpenAI-compatible API. This means it can manage most mainstream coding assistants available today.

You no longer need a separate terminal tab for each agent. Simply configure their API addresses and keys in the settings, and agent-deck presents them in a single, glanceable list. This feature is particularly useful for developers who frequently compare outputs from different models. For instance, you can ask Claude a question, then switch to Gemini to see its response, all without ever leaving your keyboard.

Real-World Use and Considerations

After a few days of hands-on use, my biggest takeaway is its 'lightness.' The interface is responsive, memory footprint is low (thanks to Go), and session switching is almost instantaneous. The history feature saved me countless up-arrow presses, allowing me to select past conversations directly from the list rather than scrolling through terminal output.

However, it does have some clear limitations. The UI is quite basic right now, lacking built-in multi-window splitting or rich text rendering. If you want to compare two model responses side-by-side within the same view, you'll still need to manually switch back and forth. Also, if you're using non-standard API ports or require proxy connections, you'll need to manually adjust the configuration file. This might be a slight hurdle for newcomers, but the documentation provides concrete examples.

Compared to tmux, agent-deck is more 'vertical'—it's not a general-purpose terminal multiplexer but specifically optimized for AI conversation scenarios. If you only occasionally use AI coding assistants, you might not fully appreciate its value. But for those who interact with multiple AI agents daily, a tool focused purely on 'session management' can genuinely boost efficiency.

Overall, agent-deck addresses a specific yet very real pain point: making the terminal workflow cleaner for multi-AI coding assistant users. It doesn't try to be everything; instead, it excels at session management. For efficiency-driven independent developers and heavy AI users, this open-source utility is definitely worth exploring.