PilotDeck

Overview

PilotDeck is an open-source AI Agent workbench and agent operating system built around the WorkSpace concept. It targets long-running, multi-project productivity work with isolated project files, memory, and skills, plus white-box memory, smart routing, and always-on background execution.

Key Features

• WorkSpace-level isolation gives each project its own file system, memory store, and skill set so parallel projects do not interfere with each other.
• White-box memory makes memory generation, extraction, storage, and retrieval visible and editable, with Dream Mode consolidation and one-click rollback.
• Smart routing detects task difficulty and sends complex calls to flagship models while routing simpler work to lighter models to reduce token spend.
• Always-on background execution lets the agent keep discovering tasks, run long-horizon monitors, and land deliverables as local files with a summary report.
• MCP-native extension and an open plugin architecture support MCP servers, custom tools and skills, lifecycle hooks, and custom memory providers.
• An out-of-the-box Web UI provides WorkSpace management, white-box memory editing, and visualization of multi-agent collaboration.

Use Cases

• Office document generation, including presentations and formal HTML white papers.
• Data analysis and spreadsheet cleanup for attendance, reimbursement, asset ledgers, and anomaly marking.
• Mini-game and app development, including an iOS AR mini-game and a low-code AI engineering platform from scratch.
• Multilingual podcast push and social media operations, including generating graphic push pages in multiple languages.

Getting Started

• Use the one-line installer for macOS or Linux, or the PowerShell one-line installer for Windows, then open the local URL printed by the command.
• For development, clone the repository, enable corepack, install with the pinned pnpm workflow, configure a model provider, and start the UI in development or production mode.
• Docker Compose is available as an alternative startup path when Docker is installed.

Deployment & Requirements

• Windows 10 or later; macOS 14 or later; Linux packages for x64 and ARM64 in DEB or RPM formats.
• The one-line installer and source workflow require Node.js 22.13 or newer and below version 23 for the built-in SQLite runtime.
• A model provider configuration is needed, such as OpenAI, Anthropic, Gemini, DeepSeek, Qwen, Kimi, MiniMax, other OpenAI-compatible endpoints, or local Ollama.
• Docker Compose can be used when Docker is installed.

Before You Adopt

• License: AGPL-3.0. Review its terms before using, modifying, or distributing the project.
• The one-line installer depends on a specific Node.js 22 runtime range for the built-in SQLite runtime.
• Demo media via Git LFS is optional; if Git LFS is unavailable or times out, installation continues without demo videos or GIFs.
• Windows source compilation may require Visual Studio Build Tools with the C++ workload and Python if prebuilt native packages are unavailable.