BOSS Console logo

BOSS Console

Open-source JVM desktop harness: BYO coding agents with browser, terminal, editor, secrets, and governed MCP tools.

Open source

BOSS Console (BOSS) is an Apache-2.0, multi-platform operator console from Risa Labs for AI coding agents. Built with Kotlin/Compose Multiplatform on the JVM (not Electron), it runs Claude Code, Codex, Gemini CLI, or OpenCode in a workspace that includes an embedded Fluck browser, shareable BossTerm, code editor, Secret Manager, Toolbox plugins with hot-reload, and 100+ built-in `mcp__boss__*` tools with per-tool RBAC and kill-switch controls. Installers ship for macOS (Universal), Windows (x64/ARM64), and Linux (AMD64/ARM64); latest public release observed was v9.2.58 (2026-07-24).

Selection advice

Choose BOSS Console when you already pick Claude Code, Codex, Gemini, or OpenCode and need an open-source desktop workspace that gives those agents a real browser, terminal, editor, secrets, and governed MCP tools. Choose OpenHands when you want the coding agent itself to be the open-source product, or E2B when the hard problem is isolated cloud execution rather than a local operator console.

Quick comparison

BOSS is a local desktop harness that hosts your chosen coding-agent CLI and exposes the workspace itself over MCP. OpenHands is a standalone software-engineering agent runtime; E2B and Stagehand focus on sandboxed or browser execution rather than a full operator console.

BOSS ConsoleOpenHandsE2B
Best forOperators who want one open-source desktop home for multiple coding-agent CLIs with governed toolsTeams that want an autonomous coding agent to edit repos and run commands in a sandboxProducts that need ephemeral cloud sandboxes for untrusted agent code
Primary surfaceNative JVM desktop app + MCP toolbox (browser, terminal, editor, secrets, plugins)Open-source coding-agent runtime with shell, files, and browser toolsCloud sandbox SDK/API for executing agent workloads
Agent modelBring your own Claude Code, Codex, Gemini, or OpenCode sessionOpenHands is the agent; you prompt and supervise its runsYou embed E2B inside your own agent or orchestration layer
TradeoffRich local console and governance; you still supply the agent CLI and model accessStrong autonomous coding loop; less of a multi-vendor operator consoleExcellent isolation for execution; not a desktop harness or MCP workspace UI

Best for

  • BYO multi-vendor coding agents in one desktop workspace
  • local-first agent ops with embedded browser and shareable terminal
  • per-tool MCP RBAC, kill-switch, and secret-scoped automation
  • JVM/native desktop preference over Electron agent IDEs

Not ideal for

  • teams that want a single bundled autonomous coding agent instead of a harness
  • pure cloud-sandbox execution without a local desktop console
  • environments that cannot run a JVM desktop application

Core concepts

agent harness / operator consolebring-your-own agent CLIMCP tool layer (`mcp__boss__*`)Fluck embedded browserBossTerm shareable terminalToolbox plugins + hot-reloadRBAC and kill-switch governanceKotlin Multiplatform / JVM runtime

Minimal implementation shape

Install with `brew install --cask boss` (or the platform installer from BossConsole-Releases), open a terminal in BOSS, start Claude Code/Codex/Gemini/OpenCode, attach the BOSS MCP endpoint (`claude mcp add --scope user --transport sse boss <url>` or the Codex/Gemini equivalents), then let the agent use built-in browser, terminal, and workspace tools under your RBAC settings.

Sources