Tessera provides an open-source workspace for AI coding agents, available as both a desktop app and a web UI. Hosted at horang-labs/tessera with 33 GitHub stars, the TypeScript project under Apache-2.0 license requires Node.js version 20.0.0 or higher. It structures CLI-based AI agents like Claude Code, Codex, and OpenCode into a visual interface for handling multiple software tasks in parallel.

The tool addresses fragmentation in AI-assisted coding workflows. Users no longer need to switch between separate command-line interfaces for each agent or model. It displays output, reasoning, tool calls, approvals, and events in real time, alongside files, diffs, branches, and pull request status. This keeps everything in one place rather than scattering attention across terminals, editors, and browsers.

Core features

Tessera organizes agent activity through several focused interfaces:

  • Parallel agent panels: Run Claude Code, Codex, OpenCode, and their model variants simultaneously. Drag-and-drop creates a multi-panel layout to view tasks, worktrees, files, and diffs without overlap.
  • Kanban board: Drag tasks across columns, generate AI-titled items, and group by collections for workflow progression.
  • Git integration: Monitors each task's worktree, branch, diff, PR state, and status updates as agents operate.
  • Composer interface: Start new panels, continue conversations, adjust reasoning, pick models, set permissions, add voice input (browser only), use @references, attach images, or send prompts with rich context.
  • Skills dashboard: Lists Claude Code skills by category, shows plugin/skill counts, enables search, favorites, and model-based analysis.

Demos in the README highlight these in action, such as multi-model workspaces across platforms and real-time Git tracking in list or board views.

Getting it running

Tessera offers options for desktop and browser use. Desktop versions target macOS and Windows in beta.

Download the latest release from https://github.com/horang-labs/tessera/releases. Extract the archive and launch the app directly—no additional installation steps listed.

For browser runtime, use the npm package at https://www.npmjs.com/package/@horang-labs/tessera. Install globally with npm install -g @horang-labs/tessera, then run it via tessera (assuming the binary is exposed). Local installation works with npx @horang-labs/tessera for testing. Ensure Node.js >=20.0.0 is present; the project badges confirm compatibility.

Both runtimes support the full workspace, including cross-platform agent parallelism. Initial setup involves configuring API keys for agents like Claude or OpenCode, though exact paths depend on the chosen models.

Report issues or ideas at https://github.com/horang-labs/tessera/issues. A design partner waitlist exists for team features like permissions and enterprise scaling.

Who this is for

Developers relying on AI coding agents benefit most. It suits those iterating from chat-based exploration to concrete tasks, such as generating code, reviewing diffs, or managing pull requests across agents.

Use it to parallelize work: one panel for Claude Code on a feature branch, another for Codex refactoring tests, with Kanban tracking progress. Solo coders or small teams handle multiple models without CLI context loss. The skills dashboard aids discovery of Claude Code extensions, streamlining plugin selection.

Voice input in the browser appeals to users preferring dictation for prompts. Git visibility helps during agent-driven development, where black-box outputs frustrate terminal-only flows.

How it compares

Tessera centralizes what separate CLIs scatter. Instead of juggling Claude Code, Codex, and OpenCode terminals, it unifies them visually. No more restarting context for production tasks or hopping tools for diffs.

Plain CLIs remain lighter for single-agent use—faster startup, no UI overhead. Browser-based agents like those in VS Code extensions offer inline editing but lack Tessera's multi-agent parallelism or Kanban. Dedicated Git GUIs (e.g., GitHub Desktop) track repos but ignore AI workflows.

At 33 stars, it's early-stage compared to mature tools like Cursor or Aider, which focus on single-model integration. Tessera's strength lies in agent orchestration; it's heavier due to the Electron-like desktop (inferred from TypeScript/web) but cross-platform from day one.

Practical limits and next steps

Beta status means rough edges persist—check issues for workarounds. It's not for non-AI coding or lightweight terminals; stick to CLIs there. Teams eyeing enterprise features should join the design partner waitlist.

Source code and binaries sit at https://github.com/horang-labs/tessera. Test the demos to gauge fit for parallel agent needs.

View on GitHub →