phodal / routa
Build Your Agent Team for Real-World AI Development - Workspace-first multi-agent coordination platform for AI development, with shared Specs, Kanban orchestration, and MCP/ACP/ A2A support across web and desktop.
View on GitHubAI Architecture Analysis
This repository is indexed by RepoMind. By analyzing phodal/routa in our AI interface, you can instantly generate complete architecture diagrams, visualize control flows, and perform automated security audits across the entire codebase.
Our Agentic Context Augmented Generation (Agentic CAG) engine loads full source files into context on-demand, avoiding the fragmentation of traditional RAG systems. Ask questions about the architecture, dependencies, or specific features to see it in action.
Repository Overview (README excerpt)
Crawler viewRouta **Your AI Agent Team, Managed by Kanban** How It Works • The Agent Team • Bring Your Own Agents • Quick Start • Architecture --- > **📦 Distribution Notice** > This project primarily provides a **Tauri desktop application** (binary distribution). > The web version is available for demo purposes only. Project links: • Releases: https://github.com/phodal/routa/releases • Docs: https://phodal.github.io/routa/ • Demo Video: https://www.bilibili.com/video/BV16CwyzUED5/ • Demo Video (YouTube): https://www.youtube.com/watch?v=spjmr_1AQLM • Contributing: CONTRIBUTING.md • Security: SECURITY.md The Idea Most AI coding tools give you one agent doing everything. Routa gives you a **team**. You describe what you want in plain language. Routa's Kanban board becomes the coordination layer — breaking your intent into cards, assigning specialized agents to each column, and flowing work from Backlog → Todo → Dev → Review → Done. Each stage has a dedicated agent that knows its job and passes work forward when it's ready. Think of it as a software team that never sleeps, where the Kanban board is both the project manager and the communication bus. How It Works • **You speak, Kanban listens** — Describe your goal in natural language. Routa decomposes it into cards on the board. • **Each column has a specialist** — Agents are bound to columns. When a card lands in their column, they pick it up automatically. • **Work flows forward** — Each agent completes its stage and moves the card to the next column. No manual handoff needed. • **Review before done** — The Review Guard agent checks implementation quality and can bounce cards back to Dev if needed. • **Full visibility** — Watch agents work in real-time. Every card shows who's working on it, what changed, and why. The Agent Team Routa ships with a set of built-in specialists, each designed for a specific stage of the development workflow: | Agent | Column | What It Does | |-------|--------|-------------| | **Backlog Refiner** | Backlog | Turns rough ideas into implementation-ready stories with clear scope and acceptance criteria | | **Todo Orchestrator** | Todo | Removes ambiguity, adds execution notes, confirms the card is ready for coding | | **Dev Crafter** | Dev | Implements the feature, runs tests, records evidence of what changed | | **Review Guard** | Review | Inspects implementation against acceptance criteria, approves or bounces back to Dev | | **Done Reporter** | Done | Writes a completion summary — what shipped and what was verified | | **Blocked Resolver** | Blocked | Triages stuck cards, clarifies blockers, routes them back into the active flow | Above the board sits the **Coordinator (Routa)** — it plans work, writes specs, delegates to specialists, and orchestrates multi-wave execution. It never writes code itself. You can also define **Custom Specialists** with their own system prompts, model tiers, and behaviors — via the Web UI, REST API, or Markdown files in . Bring Your Own Agents Routa doesn't lock you into one AI provider. Pick the backend agent that fits each task: ACP Providers (Agent Client Protocol) Routa spawns and manages agent processes through ACP. Supported out of the box: | Provider | Type | Status | |----------|------|--------| | **Claude Code** | CLI | ✅ Supported | | **OpenCode** | CLI / Docker | ✅ Supported | | **Codex** | CLI | ✅ Supported | | **Gemini CLI** | CLI | ✅ Supported | | **Kimi** | CLI | ✅ Supported | | **Augment** | CLI | ✅ Supported | | **Copilot** | CLI | ✅ Supported | ACP Agent Registry Discover and install community-contributed agents from the ACP Registry — supports , , and binary distributions. Browse the registry from Settings → Install Agents, or use the API. Multi-Protocol Support | Protocol | Purpose | |----------|---------| | **MCP** (Model Context Protocol) | Coordination tools — task delegation, messaging, notes | | **ACP** (Agent Client Protocol) | Spawns and manages agent processes | | **A2A** (Agent-to-Agent Protocol) | Federation interface for cross-platform agent communication | | **AG-UI** | Agent-generated UI protocol for rich dashboard rendering | More Features • **🔧 Custom MCP Servers** — Register user-defined MCP servers (stdio/http/sse) alongside the built-in coordination server. When an ACP agent spawns, enabled custom servers are automatically merged into its MCP configuration. • **🐙 GitHub Virtual Workspace** — Import GitHub repos as virtual workspaces for browsing and code review — no local required. Works on serverless (Vercel) via zipball download. • **📡 Scheduled Triggers** — Cron-based agent triggers for recurring tasks. • **🔗 GitHub Webhooks** — Trigger agent workflows from GitHub events (push, PR, issues). • **🧠 Memory** — Workspace-scoped memory entries that persist context across sessions. • **📊 Traces** — Browse agent execution traces, view stats, debug agent behavior. • **🎯 Skills System** — OpenCode-compatible skill discovery and dynamic loading from a community catalog. 🚀 Quick Start Desktop Application (Recommended) Web Demo (For Testing Only) Visit to access the web interface. Docker CLI (Rust) The desktop distribution includes a CLI: Community • Bug reports and feature requests: https://github.com/phodal/routa/issues • Security reports: SECURITY.md • Contribution guide: CONTRIBUTING.md 🏗 Architecture 🎯 Harness Engineering in Practice Routa is a practical case study of the three principles from Harness Engineering: build systems that are readable for AI, constrained by engineering guardrails, and improved through fast automated feedback. • **System Readability** — AGENTS.md defines coding standards, testing strategy, and Git discipline. Specialist definitions in reveal role boundaries and quality gates. Machine-friendly interfaces (MCP, ACP, A2A, REST, CLI) mean agent workflows don't depend on manual UI steps. • **Defense Mechanisms** — runs lint, delegates to . Fitness functions (docs/fitness/README.md) define hard…