• 

  SentinelOS: The Operating System for Self-Healing Software

• 

  From Incident to Verified Change Transforming production evidence into validated engineering decisions.

• 

  Autonomous Engineering Workforce Specialized AI agents collaborate to investigate incidents and coordinate engineering workflows.

• 

  Validation-First Engineering Uncertain recommendations are blocked by design and remain subject to human review.

SentinelOS

The Operating System for Self-Healing Software

Every software team knows the story.

A production incident occurs. Engineers investigate logs, inspect commits, identify root causes, write fixes, create tests, validate changes, and finally prepare a pull request.

The process is repetitive, expensive, and often happens under pressure.

We asked a simple question:

What if software could investigate its own failures before engineers even joined the incident?

That idea became SentinelOS.

What It Does

SentinelOS is an autonomous engineering platform that coordinates a workforce of specialized AI engineering agents.

When an incident occurs, SentinelOS:

• Analyzes logs
• Investigates repository changes
• Reviews failing tests
• Identifies root causes
• Generates remediation recommendations
• Creates regression tests
• Validates outputs
• Prepares engineering workflows for human review

Rather than acting as a single assistant, SentinelOS behaves like an AI engineering workforce.

How It Works

Incident
    ↓
Log Analysis
    ↓
Repository Investigation
    ↓
Test Analysis
    ↓
Root Cause Analysis
    ↓
Validation
    ↓
Human Review

Each stage is executed by specialized agents that contribute evidence to an end-to-end investigation workflow.

Features

Autonomous Engineering Workforce

SentinelOS coordinates multiple specialized agents that collaborate during incident investigations.

Evidence-Based Root Cause Analysis

The platform combines:

• Logs
• Stack traces
• Test failures
• Repository history

to produce explainable diagnoses.

Validation-First Design

Uncertain outputs are blocked by design.

Every recommendation must pass validation requirements before progressing to review.

Human-Gated Decisions

SentinelOS never removes human oversight.

The platform recommends. Engineers decide.

Incident Replay Dashboard

A real-time dashboard visualizes investigations, timelines, agent activity, evidence graphs, and engineering decisions.

Built Architecture

Backend

• Python
• FastAPI
• Pydantic
• WebSockets

Frontend

• React
• TypeScript
• React Flow
• Vite

AI Layer

• Ollama
• Qwen 2.5 Coder
• Multi-Agent Orchestration

Infrastructure

• Docker
• Git Integration
• Validation Pipelines

Challenges We Faced

Building autonomous engineering workflows introduced challenges around:

• Agent orchestration
• Context sharing
• Validation
• Trust
• Human oversight

One of the most important lessons was that reliable engineering automation requires much more than code generation. Reasoning, validation, and explainability are equally important.

What We Learned

We learned that the future of software engineering is not just AI-assisted development.

It is autonomous systems capable of investigating failures, helping engineers understand problems, and supporting safer remediation workflows.

Vision

Today's software systems generate alerts.

Tomorrow's software systems will investigate themselves.

Our vision is a future where engineering teams spend less time firefighting and more time building, while autonomous systems handle reliability workflows responsibly.

Why SentinelOS Matters

GitHub Copilot helps developers write code.

SentinelOS helps software systems investigate, repair, validate, and evolve themselves.

Built With

• agenticai
• aiagents
• automation
• devops
• docker
• fastapi
• github
• incidentresponse
• llm
• nvidia
• ollama
• opensource
• pydantic
• python
• qwen
• react
• reactflow
• reliabilityengineering
• softwareengineering
• testing
• typescript
• validation
• vite
• websockets