Scan any GitHub repository for AI-native development primitives through the hosted web app, the VS Code extension, the GitHub CLI extension, or the standalone CLI.

Is it AI-Native inspects a repository's file tree and checks for AI-native development primitives across GitHub Copilot, Claude Code, and OpenAI Codex. Every surface uses the same shared scan engine and scoring model, so browser, editor, and CLI results stay aligned.

Available surfaces:

• Web app for browser-based scans, shared reports, and WebMCP preview support
• VS Code extension in packages/vscode-extension/README.md
• GitHub CLI extension in packages/gh-extension/README.md
• Standalone CLI in packages/cli/README.md

• Repository inputs: GitHub repository URLs, owner/repo references, or local workspaces depending on the surface you use
• Assistant coverage: GitHub Copilot, Claude Code, and OpenAI Codex
• Primitive categories: Instructions, Prompts, Agents, Skills, MCP Config, and Agent Hooks
• Result shape: overall score, verdict, preferred assistant, per-assistant breakdown, and primitive-level matches

• Shared scan engine across the web app, VS Code extension, GitHub CLI extension, and standalone CLI
• Configuration-driven detection through JSON assistant and primitive definitions in docs/configuration.md
• Per-assistant scoring with an overall verdict based on the strongest assistant score
• GitHub API token support for higher rate limits on repeated remote scans
• Weekly maintenance automation that reviews assistant-related configuration drift before proposing minimal draft PRs

The web app is the primary browser-based surface. It provides the quickest path for scanning public GitHub repositories without installing editor or terminal tooling.

Web app capabilities:

• Browser-based scanning backed by the shared repository scan engine
• Shareable reports under /_/report/<uuid> with 90-day expiry when sharing is enabled
• WebMCP preview in Chromium-based browsers when about://flags/#enable-webmcp-testing is enabled
• Single-container hosting model where Express serves both the SPA and the API

WebMCP preview tools:

• scan_repository via navigator.modelContext.registerTool(...)
• scan_repository_form via the annotated repo scan form

Architecture:

┌──────────────┐        ┌────────────────┐
│   Browser    │──3000─▶│ Express App     │
│  (SPA)       │        │ SPA + API       │
└──────────────┘        └───────┬────────┘
                                │
                       ┌────────▼────────┐
                       │  GitHub API     │
                       │  (repo scan)    │
                       └────────┬────────┘
                                │
                       ┌────────▼──────────────┐
                       │  File-backed report   │
                       │  store (optional)     │
                       │  for shared reports   │
                       └───────────────────────┘

Use the component-specific READMEs for installation and usage details outside the browser:

• packages/vscode-extension/README.md
• packages/gh-extension/README.md
• packages/cli/README.md
• docs/configuration.md for assistant and primitive configuration

Scan a GitHub repository.

Request body:

{ "repo_url": "https://github.com/owner/repo", "branch": "main" }

branch is optional. The response includes the scan result plus metadata such as the scanned branch, source, and paths_scanned.

Returns server configuration flags.

Response:

{ "sharingEnabled": false }

Save a scan result for sharing when ENABLE_SHARING=true.

Request body:

{ "result": { /* scan result object */ } }

Response:

{ "id": "uuid", "url": "/_/report/uuid" }

Retrieve a shared report by ID when ENABLE_SHARING=true.

Health check endpoint returning only the runtime capability flags needed for liveness checks and lightweight diagnostics: scan token availability, report sharing status, and Application Insights telemetry status.

The repository includes a GitHub Agentic Workflow at .github/workflows/weekly-assistant-config-review.md to keep supporting skills, prompts, and scanner configuration aligned with current assistant platform documentation.

The workflow:

• fetches vendor documentation sources listed in docs/configuration.md
• compares them against the current repository-scoped assistant model
• narrows detected drift to the smallest justified change set
• creates a draft PR instead of merging directly

The web app is designed to run on Azure Container Apps on the Consumption plan.

All Azure resources are defined in infra/main.bicep with defaults in infra/main.bicepparam.

Optional deployment features:

• acrName for Azure Container Registry image pulls
• customDomainName and managedCertName for managed TLS on a custom hostname
• secondaryCustomDomainName and secondaryManagedCertName for dual-hostname migrations
• monitoringAlertEmail and monitoringUrl for low-cost availability monitoring
• enableEngagementWorkbook=false if you need to skip workbook deployment while keeping Application Insights enabled

Container scale settings are configured in infrastructure, not via the public API:

• For manual deployments, set containerStartupStrategy in infra/main.bicepparam or pass --parameters containerStartupStrategy=keep-warm when running az deployment group create.
• For the GitHub Actions production deployment, set the inline containerStartupStrategy=... override in .github/workflows/cd.yml.
• minReplicas is derived from containerStartupStrategy in infra/main.bicep.
• maxReplicas is set directly in the template.scale.maxReplicas block in infra/main.bicep.

The main Azure deployment now provisions the shared workbook automatically whenever enableAppInsights=true and enableEngagementWorkbook=true.

The workbook combines workspace-based AppPageViews and AppEvents telemetry so you can monitor promo traffic and product engagement in one place. Primary custom event names now include:

• cta_clicked_client
• landing_section_viewed_client
• outbound_doc_link_clicked_client
• scan_requested_client
• scan_succeeded_client
• scan_completed
• scan_failed
• report_created
• report_share_requested_client
• report_shared_client
• shared_report_viewed

Use the workspace-based AppPageViews and AppEvents tables for workbook queries.

To deploy the workbook by itself into an existing resource group and workspace, run:

$resourceGroup = '<resource-group>'
$workspaceId = az monitor log-analytics workspace show --resource-group $resourceGroup --workspace-name is-ai-native-logs --query id -o tsv
az deployment group create --resource-group $resourceGroup --template-file infra/workbooks/is-ai-native-monitoring.workbook.bicep --parameters workbookSourceId=$workspaceId

To deploy the full stack, including the workbook, with the repo defaults, run:

az deployment group create --resource-group <resource-group> --template-file infra/main.bicep --parameters @infra/main.bicepparam --parameters containerImage=<image> githubToken=<token>

cd webapp/backend
npm install
npm run dev

The API runs at http://localhost:3000.

cd webapp/backend
npm install
npm run dev:full

This serves the SPA from webapp/frontend/ on the same origin as the API.

docker compose up --build

Stop it with:

docker compose down

docker build -t is-ai-native .
docker run -p 3000:3000 -e NODE_ENV=production is-ai-native

Backend tests:

cd webapp/backend
npm install
npm test
npm run test:unit
npm run test:contract
npm run test:integration

Workspace tests:

npm install
npm run test:frontend
npm run test:core
npm run test:cli
npm run test:gh-extension
npm run test:vscode-extension
npm run build:cli
npm run build:vscode-extension

├── webapp/
│   ├── backend/
│   └── frontend/
├── packages/
│   ├── core/
│   ├── cli/
│   ├── gh-extension/
│   └── vscode-extension/
├── infra/
├── docs/
├── .github/
├── Dockerfile
├── docker-compose.yml
└── README.md

This project is licensed under the MIT License.