SinarisA shared control layer for AI coding agents

Turn one-off AI chats into trackable, resumable, collaborative software delivery.

Get Started

Product Architecture

Durable Project Memory

Capture goals, proposals, milestones, phase designs, tasks, and evidence outside the chat.

Agent-native Execution

Product code provides tools and state; SKILLs define how agents decide, continue, verify, and stop.

Human-visible Progress

The Hub shows plans, ready work, blocked decisions, agent sessions, activity, and evaluator confidence.

What Problem Does It Solve?

AI coding agents are strong at local implementation, but real projects break down when direction, decisions, task boundaries, and verification evidence live only in chat history.

Sinaris gives AI coding work a stable control layer:

Guide and Proposal make intent explicit.
Design turns confirmed direction into executable phase plans.
Apply lets agents claim and complete one task at a time with evidence.
Evaluator tracks whether high-value product journeys remain safe to ship.

Core Surfaces

Surface	Purpose
Plans	Durable project memory for goals, proposals, milestones, and phase designs.
Board	Claimable work, ownership, dependencies, and task progress.
Activity	Decision requests, resume messages, and agent session events.
Evaluator	Release confidence through user journeys and evidence.

30-Second Start

bash

# Install the .NET SDK first if `dotnet` is not available:
#   macOS:   brew install --cask dotnet-sdk
#   Windows: winget install Microsoft.DotNet.SDK.10

dotnet tool install --global Sinaris.Cli
sinaris --version

cd /path/to/your-project
sinaris init
sinaris hub

Then choose the right IDE entry point:

Project state	Entry point
New project or unclear direction	`/sx-guide`
Existing project has a new concrete requirement	`/sx-proposal`
Requirements need executable plans	`/sx-design`
Planned work is ready for implementation	`/sx-apply`

For Linux, non-admin installs, update commands, and resume commands, see the full Getting Started guide.