preset-io/agor

Agor - team command center for all things agentic

What it solves

Agor provides a centralized, multiplayer web workspace for managing coding agents. It eliminates the friction of running agents in isolated terminals by providing a browser-based interface where each agent session is anchored to a specific git branch with its own dedicated development environment and conversation history.

How it works

Agor acts as a management layer that drives existing coding-agent CLIs and SDKs (such as Claude Code, Codex, and Gemini) rather than providing its own model. It uses a daemon to manage a database and a process-isolated executor that spawns agents. Key architectural components include:

  • Branches as Anchors: Every task is tied to a git branch with a unique working directory.
  • Isolated Environments: One-click dev servers per branch with auto-assigned ports to prevent collisions.
  • MCP Integration: Agor exposes itself via the Model Context Protocol (MCP), allowing agents to introspect and drive the system themselves.
  • ** uma 2D Canvas**: A Figma-like board where branches are represented as cards and can be dropped into zones to trigger templated prompts.

Who it’s for

It is designed for software engineering teams who want to collaborate with AI agents in a shared, observable environment with governance, RBAC, and cost tracking, as well as solo developers seeking a more structured way to manage multiple agentic workflows.

Highlights

  • Multi-runtime support: Interchangeable use of Claude Code, Codex, Gemini, OpenCode, Copilot, and Cursor.
  • Long-lived AI teammates: Persistent agents with their own knowledge-base namespaces for durable memory.
  • Multiplayer capabilities: Real-time collaboration with live cursors, shared terminals, and spatial comments.
  • Governance and observability: Per-prompt token and cost accounting with full history and branch-scoped RBAC.
  • Progressive isolation: Multiple Unix isolation modes (simple, insulated, strict) for secure self-hosting.