Paca is an open-source, self-hosted project management platform designed to integrate AI agents as first-class team members within a Scrum environment. Unlike traditional platforms that relegate AI to peripheral chatbots or automation scripts, Paca provides a shared, real-time Scrumban board where human and AI agents collaborate on tasks, documentation, and sprint planning. Created to provide a fully customizable and free alternative to commercial tools like Jira, Trello, and ClickUp, Paca emphasizes user ownership and adaptability.

The core functionality of Paca is based on the P-A-C-A cycle: Plan, Act, Check, and Adapt. This framework enables teams to manage complex development work by allowing agents to participate in collaborative backlog refinement, execute tasks on the board, run automated verification processes, and contribute to retrospective analysis. Because agents and humans operate within the same system, every action is logged, providing full transparency into the project's progress.

Some of the key features are:

• Unified Scrumban Board: Humans and AI agents share a single real-time board for assignments, status tracking, and progress visualization.
• Built-in MCP Server: Native support for the Model Context Protocol allows for seamless integration with Claude or any MCP-compatible AI agent.
• In-App AI Chat: Project-level natural language processing turns plain English into real work items like epics, stories, and documentation.
• One-Click Revert: Every change is tracked with an activity trail featuring diffs, allowing users to roll back any modification instantly.
• BDD Co-authoring: Support for Gherkin scenarios ensures that product requirements and documentation remain contextually anchored to the codebase.
• WASM Plugin Sandbox: Extensibility through WebAssembly plugins written in languages like Go or Rust allows for secure, capability-based customization.
• Self-Hosted Infrastructure: One-command deployment via Docker allows teams to maintain full data sovereignty and privacy.

Operationally, Paca is designed for quick deployment. It runs on any Linux server with Docker support, requiring no repository cloning to initiate the installation process. Once deployed, users can invite AI agents by connecting them through the built-in MCP server. The platform treats workflows, board layouts, and fields as configurable elements, enabling teams to tailor the environment to their specific needs. By maintaining a lightweight core and relying on plugins, Paca ensures that teams only deploy the features necessary for their specific project requirements.

Some common use cases include:

• Automated Sprint Management: Utilizing AI agents to refine backlogs and draft sprint plans based on living design documentation.
• Continuous QA Verification: Employing specialized QA agents to run automated test suites and verify tasks in real-time as they move to completion.
• Natural Language Workspace Control: Using Claude Code skills to manage sprints, epics, and tasks via plain English commands directly from an editor environment.
• Secure Internal Project Tracking: Hosting sensitive project data locally while benefiting from AI-assisted task coordination and automated reporting.