Orchia lets you create, configure, and orchestrate intelligent agents powered by Claude and GitHub Copilot — with real-time monitoring, Kanban task management, and multi-agent communication.
Orchia is a local dashboard that gives you full control over a fleet of AI agents. Each agent has a defined role, a dedicated workspace, and persistent memory of your conversations.
Agents can execute shell commands, browse files, commit code, and even delegate tasks to each other through a built-in orchestration system — all within the safety boundaries you define.
From a simple chat interface to a full multi-agent pipeline — Orchia adapts to your workflow.
Connect agents to Claude (Anthropic) or GitHub Copilot. Each agent can use a different model, with its own API key and configuration.
Agents run bash commands in real-time via Git Bash on Windows. File creation, git commits, tests — everything actually executes.
Watch your agent work in real-time. Stdout and stderr stream directly into the logs tab as they happen — raw terminal output.
Each agent is scoped to a project directory. Three layers of security ensure agents never touch files outside their workspace.
Agents can create Kanban tickets and move tasks across columns (Todo → In Progress → Review → Done) automatically.
The Orchestrator role can spawn tasks and delegate to specialist agents. Inter-agent communication with security guards to prevent infinite loops.
Track token usage, command counts, task completion rates, and agent uptime across your entire fleet — all in one dashboard.
Review every file change made by a Developer agent with a side-by-side diff view. Accept, revert, or edit before committing.
Define allowed shell commands, block dangerous operations, and keep a full audit trail of every action taken by your agents.
A clean interface designed for focus and control.
Get a real-time overview of your agents — running, ready, busy — with uptime counters and current task display.
Create and manage your agent fleet. Each card shows the agent's role, status, model, and project folder.
Choose from 8 predefined roles — each with its own system prompt — or customize instructions entirely.
Watch every command, every output, every tool call stream in real-time — exactly like a terminal.
Tasks created by agents (or by you) flow through a Kanban board. Agents move cards automatically as they work.
A full-featured Kanban board where agents and you can create, move, and close tickets across workflow columns.
The Architect agent designs systems, defines patterns, and documents technical decisions directly in your codebase.
Inspect every change made by the Developer agent with an inline diff view before accepting or reverting modifications.
Track agent activity, token usage, command counts, and task completion rates across your entire fleet.
The PO agent writes user stories, grooms the backlog, and creates actionable tickets for the development team.
Configure workspace confinement, allowed commands, and audit logs to keep agents operating within safe boundaries.
Pick a role, choose a model (Claude or Copilot), and assign a project folder to scope the agent's workspace.
Start the agent, open the console, and send your instruction in natural language. The agent understands context.
The agent runs bash commands, reads and writes files, commits code — all streamed live to your logs tab.
Use the Orchestrator role to delegate tasks to specialist agents. Build a full team that works in parallel.
Each role ships with a carefully crafted system prompt — ready to use, fully customizable.