Skip to content

JacobWLMS/ARIA

Repository files navigation

ARIA

Multi-platform agentic development with musical-themed AI agents for Claude Code.

License: MIT Version Platforms

Documentation · Getting Started · Commands · Agents


What is ARIA?

ARIA brings 12 AI agent personas to your project management platform. Each agent has a distinct musical identity, specialized capabilities, and structured workflows. Instead of writing artefacts to local files, agents create and manage epics, stories, sprints, releases, and documents directly on Plane or Linear via native MCP tools.

An automated orchestrator polls platform state to determine which agent should work next, enabling fully autonomous project progression from idea to deployment.


Why ARIA?

  • Cross-phase context retrieval — any agent can pull artefacts from any phase at any time. The dev reads the architect's rationale. QA checks the PRD's acceptance criteria. Security reviews the data model. Every decision made earlier in the lifecycle is available to every agent later — no context is ever lost between phases.
  • No context switching — agents read from and write to your project management platform directly
  • Structured methodology — 5-phase scrum workflow with built-in quality gates and handoffs
  • Platform-native — uses Plane/Linear MCP tools for zero-latency integration, not REST APIs
  • Autonomous orchestration — agents hand off work to each other with full context preservation
  • Extensible — task-based abstraction layer makes adding new platforms straightforward

Features

Feature Description
38 slash commands Type /aria in Claude Code and autocomplete to any workflow
12 agent personas Musical-themed agents with distinct capabilities and communication styles
5-phase workflow Analysis → Planning → Solutioning → Implementation → Release
Multi-platform Plane (recommended) and Linear with shared core logic
Scrum practices Fibonacci estimation, velocity tracking, sprint capacity, backlog refinement
Automated orchestrator Polls platform state and dispatches agents autonomously
Cross-phase memory Any agent can retrieve artefacts from any phase — architecture informs dev, PRD informs QA, nothing is lost
Structured handoffs Mandatory context passing with decisions, questions, and artefact references
Git/GitHub integration Branches, commits, and PRs aligned with work items (optional)
Configurable autonomy Interactive, balanced, or fully autonomous modes
62 brainstorming techniques Comprehensive creative technique library for ideation

Agents

Icon Name Role Capabilities
🎵 Cadence Analyst Brainstorming, market/domain/technical research, product briefs
🎼 Maestro Product Manager PRDs, epic & story creation, requirements validation
🎨 Lyric UX Designer User research, interaction design, UI patterns
🏛️ Opus Architect Architecture decisions, tech stack selection, system design
🛡️ Forte Security Engineer Threat modeling, OWASP analysis, security audits
📊 Harmony Data Engineer ERD design, ETL/ELT pipelines, data migration
⏱️ Tempo Scrum Master Sprint planning, story preparation, retrospectives
💻 Riff Developer Story implementation, TDD, code delivery
🔍 Pitch QA Engineer Code review, test generation, quality assurance
🚀 Coda DevOps Engineer Release planning, CI/CD design, deployment strategy
🏎️ Solo Quick Flow Dev Rapid specs and implementation for one-off tasks
✍️ Verse Tech Writer Documentation, diagrams, technical explanations

Quick Start

Operating Systems

OS Status Install Method
macOS Fully supported Pipe-to-shell or clone
Linux Fully supported Pipe-to-shell or clone
Windows (WSL) Fully supported Pipe-to-shell or clone (inside WSL)
Windows (native) Supported Clone and install via Git Bash

ARIA is pure YAML + Markdown — no compilation, no native dependencies. It runs wherever Claude Code runs. The installer requires a bash-compatible shell (bash, zsh, Git Bash).

Prerequisites

Install

AI Agent Install (any AI coding tool):

Paste this into Claude Code, Cursor, Cline, Windsurf, or any AI coding tool:

Read the instructions at https://raw.githubusercontent.com/JacobWLMS/ARIA/main/agent-install.md and follow them to install ARIA into this project.

Shell install:

# Install to current directory (default)
curl -fsSL https://raw.githubusercontent.com/JacobWLMS/ARIA/main/install.sh | bash

# Install to a specific directory
curl -fsSL https://raw.githubusercontent.com/JacobWLMS/ARIA/main/install.sh | bash -s -- /path/to/project

# Install for a specific AI tool (default: claude-code)
curl -fsSL https://raw.githubusercontent.com/JacobWLMS/ARIA/main/install.sh | bash -s -- --tool cursor

Or clone and install:

git clone https://github.com/JacobWLMS/ARIA.git
cd ARIA
./install.sh                          # install to current dir
./install.sh /path/to/project         # install to specific dir
./install.sh --tool cursor            # install for Cursor
./install.sh --tool all               # install for all supported tools

The installer prompts you to select your platform (Plane or Linear), then copies the appropriate files to _aria/ in your project.

Setup

# 1. Auto-discover your team and configure workflow states
/aria-setup

# 2. Configure Git/GitHub integration (optional)
/aria-git

# 3. Get context-aware guidance on what to do next
/aria-help

Slash Commands

Type /aria in Claude Code and use autocomplete. Run /aria-help for context-aware guidance.

Phase Commands Description
Setup /aria-setup, /aria-git, /aria-doctor Platform discovery, Git config, health check
Analysis /aria-brief, /aria-brainstorm, /aria-research Product briefs, ideation, market/domain/tech research
Planning /aria-prd, /aria-epics, /aria-ux PRDs, epic & story creation, UX design
Solutioning /aria-arch, /aria-ready Architecture decisions, implementation readiness
Security /aria-threat, /aria-audit, /aria-secure Threat models, security audits, architecture review
Data /aria-data, /aria-pipeline, /aria-migrate Data modeling, ETL pipelines, migration plans
Implementation /aria-sprint, /aria-story, /aria-dev, /aria-cr, /aria-test, /aria-status Sprint planning, story prep, development, review, QA, status
Release /aria-release, /aria-cicd, /aria-deploy Release plans, CI/CD design, deployment strategy
Quick Flow /aria-quick, /aria-quick-dev Rapid spec and implementation for one-off tasks
Reviews /aria-critique, /aria-docs, /aria-write, /aria-diagram Adversarial review, documentation, diagrams
Orchestration /aria-go, /aria-dash, /aria-help Auto-dispatch, project dashboard, guidance

Full command reference →


Supported Platforms

ARIA uses a platform abstraction layer — core agents and workflows are platform-agnostic, with platform-specific task implementations for each supported platform.

Capability Plane Linear
Work items (stories) ✅ Native ✅ Native
Epics (grouping) ✅ Modules ✅ Projects
Sprints ✅ Cycles ✅ Cycles
Documents ✅ Pages ✅ Documents
Milestones/Releases ✅ Initiatives ✅ Milestones
Comments ✅ Native ✅ Native
Labels ✅ Native ✅ Native
Relations/Dependencies ✅ Native ✅ Native
Custom Properties ✅ Work Item Properties ❌ Uses labels
Time Tracking ✅ Work Logs ❌ Comment fallback
Activity Feed ✅ Activities API ❌ Polling
Intake/Triage Queue ✅ Intake ❌ Backlog fallback
Custom Work Item Types ✅ Native ❌ Uses labels
Workflow State Management create_state list_issue_statuses

Supported AI Tools

ARIA's agents, workflows, and templates are tool-agnostic. The installer generates the appropriate config for your AI coding tool.

Tool Support Install Flag Features
Claude Code Full --tool claude-code (default) 38 slash commands, orchestrator, subagent spawning
Cursor Core --tool cursor All workflows via .cursor/rules/
Windsurf Core --tool windsurf All workflows via .windsurf/rules/
Cline / Roo Code Core --tool cline All workflows via .clinerules/
Other Manual Install _aria/, reference workflow index manually

Full = orchestrator can autonomously dispatch agents via subagent spawning (Claude Code exclusive). Core = all individual agent workflows work; tell the agent which workflow to run (e.g., "Run the brainstorm workflow").


Architecture

src/
├── core/                    # Platform-agnostic
│   ├── agents/              # 12 agent YAMLs + base critical actions
│   ├── workflows/           # All workflow YAMLs + instructions
│   ├── orchestrator/        # Agent dispatch rules
│   ├── tasks/               # 17 generic task dispatchers
│   ├── data/                # Key map template, workflow manifest
│   ├── artefact-mapping.yaml
│   └── module.yaml          # Project configuration
├── platforms/
│   ├── linear/              # Linear MCP implementations
│   │   ├── platform.yaml    # Tool mapping + capabilities
│   │   ├── discovery.md     # Team/status setup
│   │   ├── tasks/           # 17 Linear-specific task files
│   │   └── orchestrator/    # Linear state reader
│   └── plane/               # Plane MCP implementations
│       ├── platform.yaml    # Tool mapping + capabilities
│       ├── discovery.md     # Workspace/project setup
│       ├── tasks/           # 17 Plane-specific task files
│       └── orchestrator/    # Activity-driven state reader
└── shared/                  # Templates, checklists, data files

How it works: Agents invoke generic task dispatchers in core/tasks/. Each dispatcher reads the _aria/.platform marker file and delegates to the corresponding implementation in platforms/{linear,plane}/tasks/. Platform-specific files contain the actual MCP tool calls.


Configuration

After installation, configure ARIA via _aria/core/module.yaml:

Setting Description Default
project_name Your project name Directory name
team_name Platform team/project Auto-discovered
platform plane or linear Set by installer
user_skill_level beginner, intermediate, expert intermediate
autonomy_level interactive, balanced, yolo Derived from skill
git_enabled Enable Git/GitHub integration true

See Configuration Reference →


Contributing

See CONTRIBUTING.md for guidelines. ARIA extends the BMAD-METHOD for multi-platform environments.

Credits

License

MIT — see LICENSE for details.

About

ARIA — Agentic Reasoning & Implementation Architecture. Linear-native agentic development with 12 musical AI agents.

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages