opcode is a desktop application for Claude Code, officially positioned as the "command center for Claude Code," with the goal of bridging the command-line tool and a visual experience to make AI-assisted development more intuitive and efficient. The project is built with Tauri 2 and provides a graphical interface. It should be noted that the official statement clearly declares that this independent developer project has no affiliation, endorsement, or sponsorship relationship with Anthropic, and Claude is a trademark of Anthropic, PBC.
[Main Capabilities] According to the official README, opcode provides the following functional modules:
- Project and Session Management
- Visual project browser: browse all Claude Code projects under ~/.claude/projects/.
- Session history: view and restore past coding sessions with full context.
- Smart search: quickly locate projects and sessions through built-in search.
- Session insights: view the first message, timestamps, and session metadata at a glance.
- CC Agents (Custom Agents)
- Custom AI agents: create specialized agents with custom system prompts and behaviors.
- Agent library: build collections of specialized agents for different tasks.
- Background execution: run agents in separate processes for non-blocking operations.
- Execution history: track all agent runs through detailed logs and performance metrics.
- Usage Analytics Dashboard
- Cost tracking: monitor Claude API usage and costs in real time.
- Token analysis: break down statistics by model, project, and time period.
- Visual charts: display usage trends and patterns.
- Data export: export usage data for reconciliation and analysis.
- MCP Server Management
- Server registry: manage Model Context Protocol servers in a centralized UI.
- Convenient configuration: add servers through the UI, or import from existing configurations.
- Connection testing: verify server connectivity before use.
- Claude Desktop import: import server configurations from Claude Desktop.
- Timeline and Checkpoints
- Session versioning: create checkpoints at any point in a coding session.
- Visual timeline: browse session history through a branching timeline.
- Instant restore: jump back to any checkpoint with one click.
- Session forking: create new branches from existing checkpoints.
- Diff viewer: view the specificbody changes.
- CLAUDE.md Management
- Built-in editor: Edit CLAUDE.md files directly within the app.
- Live preview: View Markdown rendering effects in real time.
- Project scanner: Find all CLAUDE.md files in a project.
- Syntax highlighting: Full Markdown support with syntax highlighting.
[How to Use]
- After installation, launch opcode, enter the welcome screen, and choose between CC Agents or Projects.
- On first setup, opcode automatically detects the ~/.claude directory.
- Click any project to view its sessions; each session displays the first message and timestamp, and you can directly resume a session or start a new one.
- When creating an agent: design the name, icon, and system prompt; select an available Claude model; configure file read/write and network access permissions; then you can run that agent on any project.
- For usage, you can monitor costs by model, project, and date and export data for reporting; the official usage alert feature is labeled "coming soon."
- MCP servers can be added manually or via JSON, or imported from the Claude Desktop configuration, and the connection can be tested before use.
[Building from Source] The official documentation states that Release Executables Will Be Published Soon. Before building, the following must be met:
- Operating system: Windows 10/11, macOS 11+, or Linux (Ubuntu 20.04+)
- Memory: minimum 4GB (8GB recommended)
- Storage: at least 1GB of available space
- Rust 1.70.0 or higher (installed via rustup): curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
- Bun (latest version): curl -fsSL https://bun.sh/install | bash
- Git (usually preinstalled)
- Claude Code CLI: Download and install from the official Claude website, and make sure claude is available in PATH
Build steps:
- Clone the repository: git clone https://github.com/getAsterisk/opcode.git cd opcode
- Install frontend dependencies: bun install
- Development build (with hot reload): bun run tauri dev
- Production build: bun run tauri build Build artifacts are located at: src-tauri/target/release/ on Linux/macOS/Windows
Platform-specific build options:
- Debug build (faster compilation, larger binary): bun run tauri build --debug
- macOS universal binary (Intel + Apple Silicon): bun run tauri build --target universal-apple-darwin
The official documentation also lists common troubleshooting: cargo not found (confirm Rust is installed and ~/.cargo/bin is in PATH, run source ~/.cargo/env or restart the terminal), Linux webkit2gtk not found (install the webkit2gtk development package; newer Ubuntu versions may require libwebkit2gtk-4.0-dev), Windows MSVC not found (install Visual Studio Build Tools with C++ support and restart the terminal), claude command not found (confirm Claude Code CLI is installed and in PATH; you can test with claude --version), insufficient memory during build (try cargo build -j 2 to reduce parallel jobs and close other applications).
Build artifacts include the main executable, as well as installers generated when using tauri build: .deb and .AppImage for Linux, .dmg for macOS, and .msi and .exe for Windows.
[Tech Stack]
- Frontend: React 18 + TypeScript + Vite 6
- Backend: Rust with Tauri 2
- UI framework: Tailwind CSS v4 + shadcn/ui
- Database: SQLite (via rusqlite)
- Package manager: Bun
[Privacy and Security] (official description)
- Process isolation: agents run in separate processes
- Permission control: file and network access permissions can be configured per agent
- Local storage: all data remains on the user's local machine
- No telemetry: no data is collected or tracked
- Open source: the code is open source and fully transparent
[License and Contributions] The project uses the AGPL license. The official project welcomes contributions, including bug fixes and improvements, new features and enhancements, documentation improvements, UI/UX enhancements, test coverage, and internationalization.
[Limitations and Information Not Provided] The official materials do not provide specific pricing and charging models, a specific list of supported Claude models, or operating limitations beyond platform support details; some features (such as usage alerts) are marked as coming soon. The above information that is not clearly stated in the materials is not currently provided.