Spec-driven workspace for software development with AI

Tasks and progress

Skaro keeps milestones, tasks, phases, and progress in .skaro next to the code. For each task, you can see its status, current phase, and related artifacts.

ADR

Architectural decisions are recorded as ADRs with separate statuses. This helps keep key decisions out of scattered chats and oral discussions.

Checks and tests

You can run verify commands and implementation checks for tasks. Results are stored separately so execution, validation, and manual confirmation stay distinct.

Project review

After tasks are completed, you can run a project-level review: check required artifacts, task status, and global verify commands, then start a fix cycle for detected issues.

Git in the dashboard

The dashboard includes status, diff, stage, unstage, commit, push, and branch switching. Changes produced in AI cycles can be prepared for commit right away.

skaro dashboard
Quick Start
# Install via install script
$curl -fsSL https://raw.githubusercontent.com/skarodev/skaro/main/install.sh | sh

Two commands to get started

skaro init creates the .skaro/ structure in your project, then skaro ui starts the local web dashboard.

Artifacts and working phases

Skaro adds a layer of artifacts and managed phases on top of the repository: from project rules and architecture to tasks, tests, and final review.

Constitution
Project rules: stack, standards, testing, security, constraints, and working agreements.
Architecture
System description, AI review of the architecture, and capture of architectural invariants.
ADR
Architecture Decision Records for individual decisions, with history and statuses.
DevPlan
A work plan with milestones and tasks that can be generated, updated, and confirmed.
Tasks
Tasks support Clarify → Plan → Implement → Tests, plus stage notes and related files.
Review
Project-level checks for required artifacts, task status, and global verify commands.

Git integration

View changes, inspect diffs, stage files, commit, push, and work with branches directly in the dashboard.

Status and token usage

Track project status, task progress, and token usage by phase, task, model, and role.

AI chats and fix cycles

Contextual AI chats are available for architecture, the project, tasks, and feature planning. Separate fix modes are available for tasks and the full project.

Skills and execution environment

You can connect stack-specific skills and choose where verify commands run: on the host, in a Docker Compose service, or through a command prefix.

Different models for different roles

Skaro lets you assign separate models to architect, coder, and reviewer roles to split architectural, implementation, and validation work.

Architect

Used for architecture, devplan, feature planning, and project import analysis.

Coder

Used for task implementation and fix cycles when you need fast, practical work in code.

Reviewer

Used for clarify, checks, and other phases where questions, validation, and quality control matter most.

Constitution presets for common stacks

Skaro includes ready-made constitution templates for common frontend, backend, and mobile stacks. You can use them as a starting point and adapt them to your project.

React
Vue.js
Angular
Svelte
Next.js
NestJS
file_type_django
Django
FastAPI
Express
Flutter
Kotlin
file_type_reactts
React Native
Supported LLM providers
Anthropic
DeepSeek
Google Gemini
Grok
Ollama
OpenAI
OpenRouter
Qwen
Z.ai