SudarshanAI Guides
Getting Started
Step-by-step instructions for generating blueprints, analyzing repos, and mastering every mode of the Sudarshan Engine.
Your First Blueprint in 60 Seconds
Learn how to describe your system requirements to get a complete, production-ready infrastructure blueprint. Covers prompt structure, cloud target selection, and reading the output.
- 1.Navigate to Terminal Mode
- 2.Enter your project name
- 3.Describe your system (include QPS, data requirements, region)
- 4.Click Initialize Synthesis
- 5.Review the generated blueprint, diagram, and cost estimate
Using Compiler Mode for Constraint-Driven Design
Compiler mode applies formal mathematical constraints to your architecture — deriving exact compute sizing, latency bounds, and cost ceilings from first principles using M/M/1 queueing theory.
- 1.Switch to Compiler Mode in Terminal
- 2.Enter your QPS, latency SLA, and budget constraints
- 3.Specify your cloud targets and regional requirements
- 4.Review the deterministic SKU output and simulation results
- 5.Export the specification for your IaC implementation
Scanning a GitHub Repo with Deep Scanner
Deep Scanner analyzes your existing codebase using AST traversal to reverse-engineer your actual architecture, find circular dependencies, flag security issues, and recommend improvements.
- 1.Navigate to Scanner Mode
- 2.Paste your public GitHub repository URL
- 3.Click Start Analysis
- 4.Review the architecture map and dependency graph
- 5.Examine the security findings and recommendations report
Sharing Blueprints with Your Team
Learn how to generate a public share link for any blueprint so your team, clients, or the community can view your architecture — without needing an account.
- 1.Generate a blueprint using any mode
- 2.Click the Share Link button in the output header
- 3.The public URL is automatically copied to your clipboard
- 4.Share the link — viewers can see the full diagram and spec
- 5.Links are permanent and view-only (they cannot modify your blueprint)
Migrating from Monolith to Microservices
A step-by-step prompt strategy for breaking down a monolithic architecture into domain-driven microservices using the SudarshanAI engine.
- 1.Start a new Terminal Mode session
- 2.Set prompt: 'Migrate a legacy monolithic Node.js application (5000 QPS) into 3 distinct microservices: User Auth, Product Catalog, and Order Processing'
- 3.Specify shared infrastructure: 'All services must use a shared Redis cache but isolated PostgreSQL databases'
- 4.Generate the blueprint and review the new API Gateway routing layer
- 5.Export the resulting infrastructure specification for Terraform
Designing a Global Serverless Next.js App
How to configure constraints to generate a planet-scale serverless deployment with edge caching and multi-region database failover.
- 1.Open Compiler Mode
- 2.Set high scale constraints: 25,000 Expected QPS, burst pattern: 'spiky'
- 3.Enable 'Multi-Region Failover' and set Latency Optimization Bias to true
- 4.Choose 'DynamoDB' to force a serverless persistence tier
- 5.Review the resulting Edge locations, global accelerator, and multi-region active-active database layout
Auditing AWS EKS Clusters for Security
Use Deep Scanner to analyze an infrastructure-as-code repository holding an Amazon EKS cluster configuration and uncover IAM vulnerabilities.
- 1.Navigate to Scanner Mode
- 2.Paste the GitHub URL of a repository containing Kubernetes manifests and AWS Terraform modules
- 3.Run the analysis engine
- 4.Review the generated security report paying special attention to 'Over-privileged IAM Roles' and 'Public Subnet Exposures'
- 5.Apply the suggested least-privilege IAM policies provided in the fix recommendations