MarkDiag
AI-powered runtime architecture analysis for source code, infrastructure, Markdown diagrams, and remote software systems.
Architecture clarity across code, runtime, Git, and diagrams.
MarkDiag is designed to turn scattered technical evidence into a living architecture model that engineers can inspect, export, and explain.
Interactive architecture graph
Explore services, modules, files, routes, classes, dependencies, and runtime zones in one navigable graph.
AI architecture intelligence
Generate expert reports with responsibility mapping, anti-pattern detection, risk scores, and remediation guidance.
Runtime and infra awareness
Read Docker, Compose, NGINX, systemd, env schemas, API routes, and deployment topology without exposing secrets.
Git and code context
Connect architecture nodes to branches, hotspots, commits, symbols, imports, calls, and ownership signals.
Markdown diagram parsing
Import Mermaid, D2, DOT, and Markdown architecture notes, then normalize them into editable visual models.
Read-only remote analysis
Analyze remote systems over SSH using bounded, read-only scans and explicit exclusions for production safety.
A SaaS workspace built around the architecture graph.
The desktop studio can evolve into a web workflow where teams upload projects, launch async scans, inspect the graph, and export validated architecture reports.
Runtime architecture
Generated from code, infra config, and AI analysis
Not just generated diagrams. Architecture interpretation.
The AI layer should explain responsibilities, runtime flows, weak boundaries, dependency cycles, operational risk, and next actions in language an architect can use.
Architecture report
generated after graph validation
Modular monolith evolving toward async service boundaries.
Upload, analyzer queue, AI report, graph export.
Long-running scans must never execute inside request handlers.
Discovery
code, routes, symbols, imports
Topology
services, dependencies, runtime zones
AI report
risks, bottlenecks, architecture narrative
Export
Mermaid, Markdown, PDF snapshots
Runtime intelligence
infra evidence mapped into architecture nodes
Latency
real source
CPU
Prometheus
Secrets
redacted
docker-compose.yml
web -> api:8000
api -> redis:6379
api -> postgres:5432
nginx/sites-enabled/markdiag.conf
/api/* -> fastapi upstream
systemd/markdiag-worker.service
queue=analysis priority=standardArchitecture should reflect how the system actually runs.
MarkDiag can move beyond static file trees by connecting deployment evidence, process boundaries, metrics, routes, and dependency flows.
- Dockerfile and Compose service boundaries
- NGINX upstreams, routes, and reverse proxy paths
- systemd units and long-running processes
- environment keys with sensitive value redaction
- Prometheus or JSON metrics when configured
- Git divergence, branch context, and change hotspots
SSH analysis can become a premium feature, but it must stay safe.
The SaaS roadmap should protect production systems with read-only commands, isolated workers, strict limits, and visible audit trails.
Use existing SSH trust
Rely on keys, agents, and server configuration instead of collecting server passwords.
Bounded read-only scan
Apply depth limits, file caps, exclusions, and supported extensions before reading content.
Snapshot and compare
Persist architecture snapshots locally or in SaaS storage for drift analysis.
Self-hosted deployment is the correct offer for customers that cannot send source code or topology data to a public SaaS.
Start with a focused beta offer, then expand by capability.
The clean commercial path is freemium for discovery, Pro for real analysis, and Enterprise for private deployment and controlled AI providers.
Explorer
For small projects, demos, and first architecture reviews.
- Markdown and Mermaid analysis
- Small project uploads
- Basic AI architecture summary
- Limited exports
Studio Pro
For developers and teams auditing real codebases.
- Deep project analysis
- Interactive graph workspace
- Architecture risk reports
- Snapshot history
- Mermaid and PDF exports
Enterprise
For secure teams, agencies, and self-hosted environments.
- Private deployment
- SSH and Git intelligence
- Team workspaces
- Audit logs and SSO roadmap
- Custom AI provider routing
A practical route from desktop studio to SaaS product.
The first release should prove demand, then move heavy analysis behind a resilient backend and worker architecture.
private beta
Build the architecture intelligence platform before the full SaaS.
Start with real demand, strong positioning, and a focused web surface. Then connect uploads, async workers, graph rendering, and AI reports.