Project-Nighthawk

Project Nighthawk is an AI-powered research assistant for solution engineers and cloud solution architects that generates comprehensive, fact-checked technical reports about Azure Kubernetes Service (AKS) and Azure Red Hat OpenShift (ARO).

Features

• Automated Research: Searches official GitHub repositories and Microsoft Learn
• Fact-Checked: Every claim validated against source materials
• Comprehensive Reports: Markdown files with diagrams and source citations
• Deep Technical Insights: Implementation details from actual source code
• Zero Installation: Works in VS Code with GitHub Copilot

Quick Start

/Nighthawk How does AKS implement KMS encryption with customer-managed keys?

/Nighthawk What are the networking options for ARO private clusters?

/Nighthawk How does AKS handle API server VNet integration with managed identity?

The system will automatically research, synthesize, and fact-check a complete technical report.

See USAGE.md for detailed guide.

What You Get

Each report includes:

• Overview: Direct answer to your question
• Technical Deep Dive: Implementation details from source code analysis
• Mermaid Diagrams: Visual explanations of complex concepts
• Key Findings: Summary of discoveries
• References: All sources cited (GitHub files, Microsoft Learn articles)
• Fact-Check Summary: Verification status of all claims

Repository Structure

project-nighthawk/
├── .github/
│   ├── agents/                   # Agent definitions (.agent.md files)
│   │   ├── Nighthawk-Orchestrator.agent.md
│   │   ├── Nighthawk-Classifier.agent.md
│   │   ├── Nighthawk-AKS-Researcher.agent.md
│   │   ├── Nighthawk-ARO-Researcher.agent.md
│   │   ├── Nighthawk-Synthesizer.agent.md
│   │   └── Nighthawk-FactChecker.agent.md
│   ├── skills/                   # Reusable domain knowledge loaded by agents at runtime
│   │   ├── Nighthawk-LocalRepos/ # Repo paths, git pull procedure, tool checklist
│   │   └── Nighthawk-ReportTemplates/ # Report templates, Mermaid guidelines, writing rules
│   ├── prompts/                  # Shared prompt files
│   └── copilot-instructions.md  # Workspace-level Copilot instructions
├── notes/                        # Generated research reports (output)
├── repos/                        # Locally cloned source repositories (gitignored)
│   ├── AKS/
│   ├── AgentBaker/
│   ├── ARO-RP/
│   ├── azure-cli/
│   ├── azure-cli-extensions/
│   └── cloud-provider-azure/
├── assets/                       # Images and static files
├── ARCHITECTURE-DECISION-FRAMEWORK.md
├── TESTING-CHECKLIST.md
├── USAGE.md
└── README.md

Agent definitions live in .github/agents/. Skills live in .github/skills/ and are loaded on demand at the start of each run. Generated reports land in notes/. The repos/ directory holds the locally cloned source repositories that researchers search against and is not committed to source control.

Example Reports

Check the notes/ directory for examples of generated reports.

Architecture

Nighthawk uses an orchestration framework with six specialized AI agents:

1. Orchestrator - Coordinates the workflow
2. Classifier - Determines AKS vs ARO
3. Researchers (AKS/ARO) - Deep source code analysis
4. Synthesizer - Creates comprehensive reports
5. FactChecker - Validates all claims

Each agent has specific tools and responsibilities, ensuring quality through staged validation.

The system implements the Agent Handoff Pattern where specialized agents complete distinct tasks and pass results through well-defined contracts.

See .github/agents/README.md for architecture details.

Requirements

• VS Code with GitHub Copilot
• Access to this repository
• git (for cloning and pulling local repos)
• Mermaid Chart VS Code extension (MermaidChart.vscode-mermaid-chart) — for rendering diagrams in generated reports

No additional installation required.

Setup

1. Clone local repositories

Nighthawk researchers search locally cloned repos instead of making remote API calls, which produces more accurate, grounded results. Clone the repos once before first use:

mkdir -p repos

# ARO
git clone --depth=1 https://github.com/Azure/ARO-RP.git repos/ARO-RP
git clone --depth=1 https://github.com/Azure/azure-cli.git repos/azure-cli

# AKS (optional — only needed for AKS questions)
git clone --depth=1 https://github.com/Azure/AKS.git repos/AKS
git clone --depth=1 https://github.com/Azure/AgentBaker.git repos/AgentBaker
git clone --depth=1 https://github.com/kubernetes-sigs/cloud-provider-azure.git repos/cloud-provider-azure
git clone --depth=1 https://github.com/Azure/azure-cli-extensions.git repos/azure-cli-extensions

Before each research run, pull the latest:

git -C repos/ARO-RP pull --ff-only
git -C repos/azure-cli pull --ff-only

2. Enable VS Code chat tools

Nighthawk agents require specific VS Code tools to be enabled. In the chat input bar, click the tools icon (wrench/hammer) and enable:

Tool	Purpose	Required
read_file	Read source files and reports	✅
file_search	Find files by name/pattern	✅
grep_search / Search in files	Search code across repos	✅
replace_string_in_file / Edit	Write reports and fact-check summaries	✅
semantic_search	Conceptual code search	✅

These settings persist across sessions — you only need to configure them once.

Topics Covered

• Azure Kubernetes Service (AKS)

  • Node provisioning and bootstrapping
  • Networking (CNI, load balancers, routes)
  • Security (KMS, managed identity, encryption)
  • Cluster lifecycle and upgrades

• Azure Red Hat OpenShift (ARO)

  • Resource provider implementation
  • Storage accounts and managed resources
  • Operators and controllers
  • Networking and security

Contributing

To add support for additional Azure services, create new researcher agents following the pattern in .github/agents/.

License

MIT